View Full Version : COMPUTER PURCHASE HELP: $8,000 Need to open 40Ft by 10 ft tall files in Photoshop

06-12-2007, 04:50 PM
Hi there,

They are purchasing a new computer for me at work and we would like to keep it under $10k.
The lower the better. Ive been doing research everywhere and found Boxx, Dell, Nvidia site, and a couple others. But what I need are specs for maybe a build through a local dealer.

The object of this computer build is to be able to open up a 40 foot wide by 10Ft tall file at 600DPI full color TIF or PSD in Photoshop, about 1-2GB. to open fast, and run realtime without any lag time or waiting. It doesnt matter if it is a MAC or PC, or better yet, what operating system to use? WINDOWS SERVER since it can hold up to 16Giga of memory? or the highest end MAC?

Primary programs that will be used is Photoshop and Illustrator with a Wacom Intuous tablet.

Please let me know ASAP as this will happen really fast! Thanks guys and gals for reading and any help that you can provide.


06-12-2007, 05:24 PM
Any 32-bit OS can only address 4GB of RAM - any x64 OS can access 128GB (well, for windows anyways - thats not actually the true cap of 64-bit, MS has simply capped it there).

The advantage to the Server editions of Windows is that they can access more sockets then the other variants. Make sure you don't confuse sockets with cores.

A socket = the place where the cpu is attached to the motherboard - a cpu can have multiple cores (4 at most in this point in time). Depending on the motherboard you go for, you can have anywhere from 1 to 8 sockets I believe. (I believe Tyan have an 8 socket board...) Definitly 4 anyways.

I'm not even going to hazard a guess at the specs of this computer. To open such a file 'instantly', considering the size, you'll find the limitation more in the storage then in the cpu. So, you'd either be looking into a funky RAID array....very funky for a 2GB file, or storing the file in RAM. I havn't seen any benchmarks, but Gigabytes i-RAM drives might be your ticket. RAM is volatile, so don't forget a UPS if you go that route. I think there's a way to partition RAM so that it is seen as a HDD...but with those, I believe you'd have to have the computer remain always on for it to stay that way.

06-12-2007, 05:35 PM
Currently Photoshop is only available as a 32 bit application. If you run it on Windows XP 64 bit for example it can use a maximum of 4GB of RAM. Windows Server will give you no advantage regarding memory since the 16 GB you quote require special hardware and very special software.
In your case i would contact for example BOXX and ask them to configure an optimal system. There are so many things you can do wrong when you are aiming for top performance that i wouldn't take the chance, especialy not if you have a budget that should cover it.


06-12-2007, 06:00 PM
Hi Shaun,

We have people send us their files to benchmark all the time.

If you want, you can contact me off-line, shoot me a test file simmiliar to the one in question, and I will load it up here in the lab... I can do some testing such as time-to-load, interactivity, time-to-save, time to apply a filter, etc.

We can try it on an XP32 system with 4GB and then try it under XP64 (or whatever else makes sense)

aglick@boxxtech.com (http://forums.cgsociety.org/aglick@boxxtech.com)



06-12-2007, 07:31 PM
Ill ask the obvious then, what do you need such capability for? It seems totally unrealistic.

40 foot wide = 480 inches

480 in x 600dpi = 288,000 pixels wide..... are you sure thats what you want?
72,000 pixels tall

Thats a 20,000 megapixel image

32bit rgb = 5megs per megapixel

You need a terrabyte of memory for the image, thats with 1 single layer. Your picture wont even fit on a harddrive.

06-12-2007, 07:52 PM
I have the same question as Imashimation.

40foot by 10foot sounds like Billboard advertising work... Pretty big billboard work.
But you don't need 600dpi to for a billboard, the human eye can't even see the difference between 600dpi and 300dpi on glossy paper. Just cut that file size way down and use something like the Rasterbator and upscale it; I doubt anyone is gonna be looking at a billboard ad with a magnifying glass.

If I'm wrong about this please let me know, because I'm interested to know what kind of project would need 40'x10' @600dpi

06-12-2007, 11:58 PM
I have print billboard before and they asked me 120, 100 or even 72dpi... the result was pretty fine! but it was not bigger than 8 feet...

anyway, if you have the money, go for xp64 boxx! I would also recommend a big screen like the dell 30inches... maybe solid state drive can help here for the hard drive buffer? dont know!

hehe... I feel the print company will have a big deal of troubles opening the image! :scream:

06-13-2007, 12:34 AM
I did a # of 50 X 10 ft images last year for various clients... I have a 3 year old computer... with 4 gigs of ram... etc etc nothign special for what you get today...

the key is... @ 720 DPI the image only needs to be 60 INCHES X 12 INCHES

then when printing you scale it 1000 and you get a acual size print but it is much more user friendly for you and the computer... and this worked on my computer so I can only imagine how fast it would be if I built the same comp with the new tech there is now...

oh also... I had to break it up when I got to a certain points I would save out certain layers... and bring in the flattened image and then go forward...

06-13-2007, 09:43 AM
you will need a very fast hardrive. 7200 rpm will not cut it. Perhaps one of those 15k SCSI drives will help.

OR, you could buy a solid state hardrive. I think they are available now. I am guessing that will be faster than any traditional drive.

06-13-2007, 12:46 PM
Or to put it another way, It sounds like someone who has no clue what they're doing has given you some fairytale requirement specs, or perhaps this is your first billboard job and youre just assuming they want some high quality 600dpi?

To give you some examples, a glossy magazine front cover will be 200-300dpi. The articles inside will be 150dpi.

A typical 20' x 10' billboard will usually be around 36dpi. this gives a realistic 8640x4620 image

If im asked for a 40 foot wide billboard then i'd be dropping down to 10-20dpi. Your 600dpi is living in a fantasy world

06-13-2007, 03:38 PM
Ill ask the obvious then, what do you need such capability for? It seems totally unrealistic.

You need a terrabyte of memory for the image, thats with 1 single layer. Your picture wont even fit on a harddrive.

If you had a terrabyte of memory, the question is, how would you use it? extended memory or fast storage?

06-13-2007, 05:47 PM
I've handled some pretty large images, but those image specs are insane. I know no one in Hollywood would bother generating images that large even with a huge database backbone with bluearcs.

If its for a billboard on the street, its overkill, you're wasting resolution on image detail that will never be seen. As you increase viewing distance, you can drop the resolution without anyone noticing.

I know guys who do vinyl wraps on H1 Hummers using 90 dpi graphics and get the printer/cutter to scale things up in hardware and the results are nice. And they're not even printing at 1:1 scale on the truck, their image was based on a 4k square.

Now if its for an art project in a museum that will be viewed close up, that may be different, one of my co-workers did a ultra high resolution photographic print that was shot with a custom camera setup that far exceeds any commercially available camera and even then he wasn't dealing with terabyte sized images.

To handle files that are that big, you'll need infrastructure that will far surpass $10,000.


Of course if the image is vector graphics, then you can get away with some relatively small files.

/edit #2


I'd read up on the bottom part of this article. The larger the image, and the greater the viewing distance, you can use much lower "dpi".

06-13-2007, 06:14 PM
working with prints for advertising every day i have to agree with the posts above. billboard files are normally produced in scale 1:10, hence often the optimal resolution is 600 dpi in scale 1:10 (which is 60 dpi in scale 1:1). anything from 36-72 dpi is considered fine for billboards that you would look at from a distance that is any larger than a meter or two (normally you cannot see the whole thing from a smaller distance than that).

06-13-2007, 06:24 PM
Alright, this may make more sense to you all.
I work for a laser etching company which has one of the biggest lasers in the Northern Hemisphere for what we do which is primarilly laser etching marble, granite, glass, travertine, you name it, with art and photography. We just got done building a really fast computer for a job for a giant casino door panel measuring 20 Ft high by 10 Ft wide, or something around those dimentions, it was huge to say the least and the new computer we had lagged big time under photoshop, we spent around 4k on that system which we can probably upgrade to a faster Quadro Card, and maybe Dual Xeon CPUS, the thing is we still dont think that will be powerful enough to do other jobs which are coming up. So they are investing in a workhorse system and really dont know what route to take yet.

To give you an idea of the capabilities of this laser, it has a working bed of 6FT by 10FT and can shoot up to 1200DPI. The jobs we shoot require the uttmost detail especially since they are being shot into a very hard subsurface, not like canvas or paper. It is not billboard printing which you can have 72DPI in some cases. The minimum DPI we run is 200DPI, but in most cases do 300DPI as well.

The bigger jobs will be shot with 200DPI unless required by the client to shoot at a high resolution for their images or original art to bring out more detail upclose. To give you an example of what can be done. If we had a working file of the Mona Lisa which in Photoshop was created at 80FT high by 30FT wide, and everything got refined so when you zoom in you can see every wrinkle perfectly without blur, our laser can shoot that exactly from the computer in those dimentions, It would obvioulsy require prolly 30 10Ft wide by 10FT tall slabs of marble, but it can be done, and when you see it to scale, its breath taking.

The job this computer is being built for is for a backpanel of a pool for multi million dollar spec homes. The Piece required to go to print is 40FT wide and 10FT tall. I will be creating the giant matte painting of a mission inside of a lush colorful garden and valley and spectacular sky. I need a computer to be able to handle such a file so I can work in my details throughout the piece. It cant be fuzzy or upsized as the client wants to see the detail since hes investing a lot of money into the project. Depending on how the first one goes will depend on how many more of these jobs we get.

With all that said, I hope this makes more sense. There are so many options out there that we just want the right one that will work in this case, "Photoshop=Huge File size=streamline workflow=minimal lag time". Thats it..

Thanks for your help thus far, thats why I came here cous I know you guys know your stuff and maybe have done similar jobs throughout your career.


P.S. excuse me if any of this doesnt make sense or if im vague, im at work and need to write this fast

06-13-2007, 06:54 PM
Wow, i think this is the first time i ever heard of a reasonable use of such extreme image sizes.
To the problem at hand. I doubt very much that Photoshop will be able to handle such a file at all, regardless what hardware you throw at it. This sounds like you would need a tailor made clustering solution and i'm mostly talking software here. Even if you were to put 1 TB of memory in a single system the bus and IO speed would cripple it.
Where do you get the original data from? What system and software were used to create it? I guess the filesize of 1-2 GB you mentioned is a compressed image format?


06-13-2007, 07:19 PM
Wow. Things make a lot more sense now.
I'd say give the BOXX guys a call... or IBM for one of their cluster setups... or NASA :D . Thats a lot of data-processing.

06-13-2007, 07:21 PM
It sounds very much like you need a custom fit imaging program. Photoshop is only 32bit, and as such is limited in the amount of data it can process. On top of that I don't know how well it utilizes multiple processors/cores. Photoshop is not ideal in this situation.

A few ideas on speeding things up a bit, not sure if this will help much though:

1) Get an extremely fast RAID array for photoshop's scratch disk. If possible using 15krpm sas disks, or better yet high end solid state drives. The faster the better, as hard disks are generally an order of magnitude slower than RAM, and your scratch disk is a sort of way of cheating Photoshop into using more RAM than what is available to a 32bit program.

A multidisk RAID 0 array with 15krpm SAS drives comes to mind....

2) Ensure that this fast RAID is hooked up through an interconnect that can handle the throughput. PCIe comes to mind, though it may not ideal...
3) Use a 64bit OS and 64bit software. This will raise your RAM cieling a bit, and hopefully give enough of a buffer that you will be able to interact with images more quickly. Granted PS doesn't have a 64bit version to my knowledge, which is really too bad. Its one of the apps that could benefit a good deal from it..
4) Top of the line quad core CPUs are probably good to have. A pair perhaps. I don't know of Photoshop's multi-threaded-ness though, so that might not help much, but it will at least be able to utilize some of that CPU power for disk I/O which you will need alot of :)
5) I don't expect an upgraded video card will have much impact on performance with your photoshop experience. Though, if you do any 3D I'd expect it to help some...

06-13-2007, 07:39 PM
what about running the GIMP (a linux based Photoshop alternative) in 64bit Linux?

This way you could put 64GB or more in a system and have aprogram that was able to load and work with HUGE datasets such as this?

Why Photoshop? Why NOT GIMP?

BOXX labs

06-13-2007, 08:15 PM
I was going to recommend GIMP, but I didn't find a 64bit Windows build of it.

I suppose though, if you're good enough, you could always roll your own ;)

06-13-2007, 08:56 PM
Maybe give this a shot on a boxx machine


64-bit of course

06-13-2007, 11:19 PM
I read SEL's original post and was WTF (just like everyone else).

Then I read SEL's reply, and while I can see him needing a powerful PC, I still think something is way overkill here. OK, so you are going to be printing (or etching) on marble, glass and other such surfaces. And this machine has a resolution of 1200DPI.... ok, well lets look at this realistically here. 1200 DPI is less than .001" - a piece of copy paper is about .005" thick, meaning that you are etcing lines 1/5th the thickness of paper. That isn't totaly unheard of, except that marble, granite and such materials are fairly brittle and fairly rough. I couldn't imagine getting anywhere near that resolution on those materials. These are resolutions so small, that you honestly wouldn't even see them.

Also, having been both in the printing industry and the manufacturing industry which uses some truely high precision, I just don't see where or why you would need something that high-rez for this project. Professional Printers (presses) might be able to print things out in 600 or 1200dpi (or more), but the input file is never that high - 300dpi at most, and that's when printing on nice smooth, sharp paper.

I would be willing to bet that feeding this machine a 100 DPI file will look EXACTLY the same as feeding it a 200 DPI file... except that the 200DPI file will be four times the size.

Have I ever worked in this granite industry? No - but from my experience in other industries tells me that someone is taking over-hyped marketing specs a little too seriously.

This situtation reminds me a little too much of the time a few years ago when my non-technical buddy was savings files out at 720 DPI. When I asked him why, he replied because his new Epson printer has a resolution of 720 DPI, and he wanted every dot to corispond to an individual pixel. It took a while to explain to him that it simply doesn't work that way.... eventually I had to show him that a 200 DPI file was printing exact the same as the 720 DPI files that he was making - but it was printing faster and using a fraction of the RAM and disk space.

06-14-2007, 02:24 AM
Laser data. Hmm.

Does the laser control software support solid model data? Instead of applying, a gi-normous image map through photoshop? or is the laser/CNC control software limited to only taking images as input?

I'm wondering if engineering/industrial design modelling software could be of some use here?

I'm in agreement with one of the above posters about the depth quality of the image, how many levels does the laser support? 8bit images? [256 levels?] or can it support 16 or 24 bit images for depth data?

06-14-2007, 01:17 PM
Your laser may very well do 300, 600 or 1200dpi, but that doesnt change the fact that what youre asking for is far far far off in the realms of impossibility. The computer you're asking for doesn't exist and won't exist this decade.

The best machine youll find will have 16 memory slots; the most dense ram youll find is 4gigs, so you could get a 64 gig computer, but thats still 936 gigs short of your terrabyte. If you want an undo buffer youre probably about 2000gigs short.

You have 3 options I can see:

1) Crank down the dpi twenty fold
2) Split the whole thing into 100 smaller segments, but good luck editing that
3) RTFM on the laser etcher and find the interpolation options. I've worked with large scale industrial lasers before, used with 3 inch thick acrylic, all of these would quite happily take in a bitmap image and then interpolate between the pixels so you dont get a large scale pacman image

06-14-2007, 02:24 PM
I would say that software might be the biggest problem here. I dont see a problem in hardware, RAM is a buffer, HD is streaming data at 1gb+, unzooming uses LOD to ask only every Nth pixel etc. Also as someone said its most prolly a 8 bit image or 16bit. As for similar software, google earth comes to my mind.

It most prolly exists, as for getting it, it might be a problem...

06-14-2007, 02:38 PM
The best machine youll find will have 16 memory slots; the most dense ram youll find is 4gigs, so you could get a 64 gig computer, but thats still 936 gigs short of your terrabyte. If you want an undo buffer youre probably about 2000gigs short.
Well not entirely true. 4-Way Opteron boards support up to 128GB of RAM.

Anyway.. I think to make the best of it a very fast RAID array for PS (and probably Windows') scratchdisk/virtual memory is probably a good way to start to address the problem. If you can get the I/O fast enough on the scratch disks, it should allow photoshop to store/retrieve data at a more reasonable rate.

I recall seeing benchmarks of Gigabyte's iDrive. Its basically a PCI card with a bunch of RAM on it (up to 2 GB I think). This hooked up to your SATA port and allowed you to use the RAM as a hard disk. In the benchmarks they compared adding an iDrive to just simply adding more RAM, and found the performance difference to be fairly minimal. Naturally, its better to use more RAM, but when your app can't address it all, you need to take short cuts.

Getting an extremely fast RAID array can do this. Granted it will always be slower than actual RAM. But the point here is that its faster than not doing it :)

06-14-2007, 02:57 PM
I'm on the list for Tyan PSC, I guess this might be a hardware solution for you. You just need software that can use it...


06-15-2007, 06:47 AM
Hi all!

to get back to the topic at hand especially to "SZOS & Imationation", I would not bring this subject to CGTALK readers unless I knew what the hell I was talking about and what this machine does and can do. Your comments about impossibilities and the fact of things not being capable to achieve is nonsense in the idea that your blowing it completelly out of proportion, please read on. The mear fact is we shoot 6X10 Foot Tiled images into tiled stone everyday. We do Acid Deap Stone Etching too that after a laser shoots the image into stone, our local scientist as we like to call him takes the laser engraved image and acid etches the stone to create depth into it. What the industry has used for years is Sandblasting which is not what we do, we can get the finest detail whereas sandblasting would not achieve even close results to some of these techniques though has a beautiful feel and look in its own right.

As expressed on my last post, I myself created a mocup of a dragon and a phoenix the size of 18 Ft High 10FT wide, in extreme detail, meaning every inch of it is highly detailed to the eye. To explain what this laser does, it takes a necessary 200DPI image, and shoots pulsating lasers shots (dont know if thats the right word) in extremelly high accuracy and speed into any subsurface, almost that is as we cannot do PCP Pipe because of the toxic chemical fumes that the laser creates when blasting that material. And transfers whats created in the computer onto the material.

We are currently one of the first if not the first to dive into a color process that will streamline colorization onto many different subsurfaces with close to the same accuracy seen on the computer. Today, most is painted by hand with either airbrush or oil. Though we are in the process of maybe purchasing a computer airbrush system that will colorize into the etching of the stone so it takes color perfectly within a CMYK spectrum, though the color process we have developed right now can do 7, lots of testing has been going on to achieve the look we have been striving for. To give you an example. Our goal by the end of the year is to be able to Laser etch something like the Mona Lisa 15X25 FT high into the marble with exact color representation that can be achieved into a dark stone. Marble properties does not react well with the acids we use for to create our depth unless it is white marble, so only the topsurface would be etched (blasted) by the laser. Our problems right now are the fact that it is stone and not white paper or canvas, so it is a more muted look with the color process, and we are looking for more pop, which might deal with the type of materials we use to color it with which weve used just about everything.

To give you an example of how DPI works with the laser, we have run countless tests on many different surface and sizes and DPI ratios to see what works best and what doesnt work at all. 100DPI on a 12X12 Surface you can see the individual points shot, the clarity is nice but not desired, we shoot most with 200DPI setting, never under that as unless you work with it everyday you would never understand why. Our lab specifys us to not go under 200DPI as well, we do sometimes on occasion for colorization test which is a whole new ballgame.

In the matter of using 600DPI and 1200DPI, we would rarelly use those settings unless the job required them. A perfect example of when we would use that is when you are trying to achieve extreme detail where you dont want to see etching points. Lets say its a portrait of George Washington that will be printed on a 6X10FOOT piece of Marble, or slabs of marble. Well, lets say the image, you can zoom into George washingtons button on the computer an get a fine detail of the button on his coat, well in this case we want to see that detail on the surface when the laser shoots, because it is so big we have the chance to shoot maybe 600DPI or higher. What that will give us is an obvious higher Dots Per Inch etch which may blow out the stone a litte, but you lose the Pixelation and gets amazing results. We tried tests of shooting 600DPI on 5X7inch tile of a photograph and it just blows it out, mainly do to the fact that its 5X7 inch size shooting so many dots per inch in a small area, and since the file size is smaller, you have a lower quality image being etched with high DPI, the image is perfectly visible but its evident of the stone being more blown out plus taking 3times longer to etch. It does not work the same as your everyday color printer incase anyone is wondering. Because it is an actual pulsating laser that will shoot through your hand if you leave it there, it requires all art or photographs to be converted to GREYSCALE and then into an 8bit BITMAP, from there it is brought into other programs developed for the laser which takes the 8bit bitmap file and converts it to an optimized 1bit bitmap file, 0/1 black white, white being laser etched, black being transparent in laser terms, the DPI plays a key roll in the fact of the amount of dots now being etched in a limited surface area. There is a lot of information that goes with this. There are many different processes that are used with different surface, different lense settings, datum and grid points, and intensity and speed factors that play a roll also in image quality. With that said, everything that I brought up here is meant for a reason and anythingg can be achieved, its pretty amazing when I first saw it work in action, and its even more amazing what its can do with wood, breathtaking actually.

The info. I got so far from this thread has been extremelly helpful minus a couple posts that have no clue what they are talking about unless they do this as their trade day in day out, so please leave the NASA comments and having moron friends who think they know everything but have no clue to a minimum. I am the art director of this company, and need to know what I am doing and know the system and everything involved to get the best images to our clients possible. We are a highly funded company and required to know everything and if we do not, are required to find the answer. In this case, the computer!

The Reason for this thread is to get the best knowledge from people who run high end computers and take it from there. I am seeing Photoshop as most likelly the main problem now, I will look into other software packages like GIMP which I dont know nothing about but may end up going that route with a Linux Setup. I dont know, but will find out shortly.

In the mean time, I will end up having to work with a downsized file in Photoshop, and then chopping it up into 4 10FT pieces and work in detail from there, until this computer gets built. Thanks for all the help to you guys, Ill most likelly give Adam a ring from Boxx to see if they have some options.



P.S. Lastly, I did get a PM regarding more .info about what we do and are capable of doing. So if your an Interior Designer, Landscape Architect, or Commerical or High End Home Architect or Contractor or know someone that works in those areas of business that are looking for something new and exciting to sell or add to their business, we would love to hear from you..

06-15-2007, 10:29 AM
You asked for something far off in the realm of fantasy, if you had mentioned at the start that all of this is going to be reduced to a 1-bit image, thats one hell of a difference. We can only comment on what youve asked for, the difference in computational power needed for an image with 16 million colours as opposed to two colours is immense.

Please dont take offense when we ask if the person providing you the requirements knows what theyre doing. A lot of us work with clueless technically incompetent clients, bosses and co-workers every day. Being given artworks specs in meaningless measurements, formats and misunderstood terminology is a daily occurance.

If you ask some people how big they want their billboard poster, youll get idiotic replies like "100 megs should do" If you ask how big the product viz printouts should be, someone will tell you its gotta be 4000 dpi. If you ask an advertising agency if the video is for standard def tv, hdtv or cinema, theyll tell you to just make it as big as it goes, theyll scale it down later.

So, when I ask if the person asking you for a 600dpi 40x10 foot image knows what theyre asking for... especially given the fact that the file would be a terrabyte, its not unreasonable to assume someone doesnt know what theyre doing. In this case it was just the missing info that its 2 colours, not 16 million. Instead of needing 1,000 gigs, it now only needs 1.7 gigs, so we can now start asking some useful questions about why your current machine can't handle it.

So lets start with the basics, what machine are you currently using, we dont want to just re-recommend the same computer youre already using. Next up, which os are you using, and if its windows 32bit, have you enabled the /3g boot.ini option?

Windows 2000/xp 32bit by default can only use ~1.7 gigs of ram per program, anything more than this and photoshop will be using its own internal scratch disk on your harddrive. If you edit your boot.ini file to include the /3g switch, this will let photoshop use ~2.7 gigs of ram. As your tiff files are 1.5 gigs, this alone could be enough to fix your slow speed problems. Alternatively windows xp 64 (or vista if youre masochistic) will let your 32bit version of photoshop use 4 gigs of ram.

Graphics cards, a quadro will not speed up 2d photoshop work, dont let any workstation salesmen tell you otherwise. All that matters is pixel screen fill speed and frankly any gfx card can display everything photoshop has to offer without any slow-down.

Personally for the work you're doing I'd grab a quad core intel core2, pack out the machine with 8 gigs ram, run it off a couple of 15k rpm drives in a raid0, no fancy gfx card and finish it off with a 30" 2560x1600 screen so you can see what youre doing.

But then again im a rude bastard so you probably dont want to listen to me.

06-15-2007, 03:25 PM
Thanks for your reply as I expected one from you. There was plenty of great info. provided including some in your last post. The bottom line is, if you read what I wrote thouroughly, I may have not been as thorough as I like. "IM NOT WORKING WITH A 1BIT IMAGE DURING CREATION!"

Lets get that straight, im working with the fullcolor spectrum, RGB, 200DPI File size minimum files. The file size will be a full color 200-300DPI file. It has to be laser wise, I dont understand how you dont get that. I dont know how you work, but its nearly impossible to paint highly detailed scenes with 1bit bitmap black and white files? especially since there are no layers. The scene needs to be created the scale of the subsurface its printed on, and the DPI is changed there depending on the detail and size. Its really simple in that aspect. Once The beautiful scene is finished and there is detail in every inch of the color photoshop file. It is then preped for the laser which is a whole nother ball game. Our color process deals with separation of colors as well in the file state transfered to the laser to.

Simple fact is,
A: Im working with a high res, full color, RGB, or CMYK file with a minimum of 200DPI in large scale.
B: I need a computer to run that big of a file.
C: Suggestions

I got more then enough here to know what routes I have and will take, thanks to all for your help, much appreciated..


06-15-2007, 03:55 PM
So this is what I understand:

1) Create image in PS at that huge resolution with high DPI (200 or higher)
2) Drop the file to a grey scale image (8bit)
3) Plug grey scale image into specialized software for the laser to create the 1bit image (which I assume is to tell the laser where to draw and where to not draw, a sort of mask)
4) The laser takes the 8bit grey scale image into account (I guess for laser intensity) while working in its masked area

Is that about right?

06-15-2007, 07:25 PM
Let us know if you try Sartori. It promises a lot and I am curious as to how well it actually works. Esp. in a work area such as yours.

06-15-2007, 09:56 PM
There are a lot of very knowledgable people here that have years/decades worth of graphics and hardware experience. If you are not getting the info that you want to hear, then I don't understand why you don't go back to the manufacturer of this laser etching device and ask them what they recommend.

I have many years of experience programing CNC machines using CAD models which is not unlike the device that you mention (yes, I know they are not the exact same, but computationally, they are doing somethign similiar) - not a one of those machines needed anythign super ultra high tech to process the data and run the machine to create the part. I am sorry, but I simply do not buy it when you mention some of the requirements of this machine. If the original artwork needs to be 24-bit, but then it gets processed down to 1-bit, then you would simply not be starting with the same size as what the machine outputs. Sounds like this convertion process is basically making a 1-bit half-tone of the color image. If that is teh case, I do not see why a near super-computer specs are being called out for.

Also, at this stage I would like to query as to why vector artwork isn't being used. Vector art can be scaled up or down to whatever size is required, yet would take up very little space - and bitmap images can be converted to vector art.

06-16-2007, 01:28 AM
Szos -Ignorance is bliss!!

Its like you dont read or something, everyone else was very helpful except yourself and imashination, though at least he tried to help spec wise whereas you are just completelly ignorant. Go spend Half a million dollars and buy one and you tell me how the hell this thing runs since you know so much! While we do vector art as well, please show me your artistic background of creating a realistic image through Illustrator, thats great because it would take you a million years splicing things to get the same detail Photoshop provides as matte paintings or realistic photographs. I will be taking others advice in purchasing other software for this that can run under a 64bit system.

"There are a lot of very knowledgable people here that have years/decades worth of graphics and hardware experience. If you are not getting the info that you want to hear, then I don't understand why you don't go back to the manufacturer of this laser etching device and ask them what they recommend. "

---> If you read I got all the info. I needed and thanked those that helped out with my questions. There are always a couple CGtalkers that love to just BS or act like they are gods gift to knowledge, or art, or something special as yourself, dont bring others into the mix as they answered my questions or helped by giving advice or work arounds or ideas. Your just wasting my time, I havent heard one helpful thing that you said. There are only 5 other lasers this size ever built for our particular industry. The manufacturer only builds the stuff and give recommendations, they are not artists, there is many uses for this machine. Today we etched an exact replica of a villa matte painting 3X9 Ft onto black granite.

"I am sorry, but I simply do not buy it when you mention some of the requirements of this machine. If the original artwork needs to be 24-bit, but then it gets processed down to 1-bit, then you would simply not be starting with the same size as what the machine outputs."

--->Eureka, youve solved the puzzle of the missing link! Dont buy it then and go on with your life of ignorance, maybe you just dont want to understand it or you are failing to. I would not be bringing up this topic here if I had a solution, ive been a part of this community since day one and know of the knowledge that flys through this site, thanks for your highjacking of the thread, Good day! Anytime you like, you can come to our studios and see the process for yourself, ide be delighted to sit you through everything and walk you through our site....

"Lots" - You hit it right on the dot. That is the process for Regular images in the most basic sense. There is another process with the laser where you need to change the head so that is can shoot vectors, its a high intensity and can slice right through things, but also create amazing 3d imagery in thick wood and acrylic that provides depth based of gradients. I wish SZOS could just understand that simple of a process. The final image, now saved out of photoshop and brought into another program is broken down to a laser ready 1bit file for the laser to process points to know where to shoot.

slatr- Thanks, I will be looking into that and pass it on to my bosses.

KurtW - Ive used maya and other 3d software for years, while the laser can use CAD software in its vector mode to create depth in wood, that is not what its being needed for now. You hit it on the dot in its limitations. Unfortunatelly, that is how it was designed and built, and requires an image the size of the subsurface its being shot onto. The DPI will shoot more or less laser etched points within a one inch ratio, though the higher DPI takes longer, on bigger images, it shoots more realistic as you tend to see the points shot if you go under 200DPI which is a bad thing in the sense that it would be like looking at a giant image of pointalism. I eventually will be creating a thread regarding this fairly new technology and show examples of things that can be done and ways its used.


P.S. Szos - dont bother replying back as I wont answer anymore of your questions unless you would like to visit the facility yourself and see it first hand what we do.

06-16-2007, 05:47 AM
I've seen you trash so much good advice, its quite shocking really. I especially have to laugh at how you've dismissed the vector route when its quite clearly the only one which can handle the dimensions you need.

Anyway here,

This is the only option I see for you if youre determined to brute force your way through. Work at 10-20dpi then run it through this as the final step. It vectorises the image, rescales it in vector format then re-rasterizes it

06-16-2007, 11:14 AM
LOL !!

Wow, why am I not surprised at the outcome of this thread.
You start up a thread with fairly vague description of your problem, have a preposterous hardware requirement, and when everyone questions the ludacris specs, you balk at the replies.

You obviously have no idea what you are talking about - you might think you do, but you obviously don't.... and some of us are sick of seeing that time and time again at some of the companies we work with. Seemingly knowledgable people that should know how to do their job, but don't.
No one, and I mean absolutely no one, is going to sell a piece of equipment to do a task as mundane as etching marble, and require supercomputer-level hardware to run it.

My final suggestion is to call NASA and see if you can borrow some of their CPU time. :rolleyes:

06-16-2007, 03:35 PM
Just wondering - how do you deal with registration between the various slabs of marble/substrate that make up the whole image? When its assembled together on site I mean? What sheet size can your table cut?


06-17-2007, 02:37 AM
I think the fact of the matter is that if you cant get someone to write you a custom software solution then you will have to make the image and the data you send to the laser is small handelable chunks. The hard ware although esoteric and pricey is'nt realy a problem, although the budget might be, run it by boxx or look at some of Tyan's desktop super computers.

06-17-2007, 03:45 AM
Heh, well calling names isn't going to help anyone :P On either side :P

Since a machine like this sounds like it costs a fair bit of money, I assume that you guys have a team that has experimented to find the optimum way of creating high quality images at minimal computing effort? If not I suggest you guys do :P You wont be able to use the machine quite as effectively if you don't know its limits :P

I'm not going to pretend I know anything about high resolution print or marble etching, but I have to agree with everyone, that is a huge amount of data to deal with.

Short of solid state drives, what I said should help to a point. But I wonder. There is probably a smarter way to approach this problem rather than brute force.

For example, you could create smaller images then stitch them together into one big image before converting them to the final image for printing. It would definately take some planning, but ultimately it should make dealing with the file size a bit more feasable. Like mentioned above. Probably one of the only ways to deal with the image sizes you mention.

06-17-2007, 09:41 PM
imashination - Despite your Laughter towards vectoring. Atleast your providing some helpful info. towards this discussion. Unless you work with this machine each and everyday, you wont understand it completelly unless you were trained from the bottom up. Please provide the quotes"" of where someone helping me on this thread I trashed their breakthrough info. as I have not, those that replyed with help or insight or workarounds or specs, all info. was recorded and brought to attention with the other staff, the bottom line here is this. You have not worked with this machine or software, so how could you possibly know the ins or outs or have any idea of where to go knowledge wise, you brought up key points, and if we do need to do a chop job, then we may just have to do that, but I dont think we have to based off of what I got in this thread, I will be talking with boxx Monday and most likelly be purchasing some new software to boot. Your Vector Comments while justified from an outsiders point of view sound good on paper, but will just not work for this machine, KURTW brought up a big point in the fact that its build gave limitations, meaning, if the size of the slab is 6X10FT, the image dimentions of the file needs to be 6X10FT. 10-20DPI just wont work even though it sounds valid to do so. A minimum of 200DPI is needed when bringing it into the software for the machine. The DPI is key to the detail process of how many laser pulse shots are shot within an inch perameter. If we shot only 10-20DPI, there would not be enough etch marks to see detail. Remember this is shot a lot of times on black marble. So a higher DPI is required otherwise you would just see black with a few tiny holes. We have done testing with DPI and file size from 10DPI all the way to 1200DPI, with files with extreme detail. We have spent months with all this, and I must tell you, ive only scratched the surface of what the machine can do and requires etc. The vector is great file wise because of the ability to size up infinity, but its not great software wise in the fact that its like telling a matte painter who does Lord of the rings backsplashes to do it in illustrator with vectors. It just is not logical and would not happen, ask any matte painter and they will tell you the same. I guess in the simplest of forms, imagine your doing a photoshop painting 6X10FT, starting from your bottom left, you draw the grass perfectly, hard edged, you have little bugs flying around, you have mini rocks and water puddles, etc. This machine will shoot that detail in those dimentions onto any subsurface. We already have shot things to that size, our problem now we are facing is a hardware softawre limitation of doing something 40FT wide by 10FT tall. As of right now I need to work at 1/4 the size in order be able to create the matte painting, if I did not have the ideal machine, I would need to then create 4 images after I get as much detail in that initial file size. I would then have to start detailing in each file and hope they match as close as possible. Ide rather not do that but its necessary when a client wants to see the little lasy bug at the bottom left corner, it cant be blurry from upsizing it. These are not billboards, they are seen upclose, and every inch must be visible. Thats the nature of the beast when we get paid a lot of money for those reasons. PERIOD. I think its going to come down to a little of everything, finding the best machine to run the software, finding a combination of software to work with the file, and finding little workarounds to speed up production time. Thanks for the link to the ononesoftware, as that can maybe be used too.

Szos - Why am I not surprised in general that you would write back with yet another comment wasted post. Open your eyes, there are reasons why people spend almost half a million dollars on a silly scanner. Huh, why does it costs 4X more then the finest printer. Ask yourself why do people even built such silly machines. I mean, why even bother with technology in general. Your posts have no solid concrete information or backbone except the fact that you must continue to highjack a thread you dont understand. Read through it again and you may get a better understanding. Its funny reading your post and seeing your missing the most key part to all this which baffles me in the fact that it doesnt take a super computer to run the laser, its going to though take a combination of hardware software and workarounds though to get the initial file ready to send to the machine to etch. "You obviously have no idea what you are talking about" - Dont quit your day job, guys like you are a dime a dozen, ive run into many of your type in the industry and yet it always seems you guys get nowhere career wise and have to bock through an internet portal to make your career choice or maybe lack there of sufficient enough to go on living. Im still waiting for a post of your great knowledge to help in running a rather large file 200DPI full color. If you have no idea, which im sure you dont, then go bothers someone elses thread.......:thumbsup:

SteveB - Hey Steve, the table is 6X10 ft. It all depends on the substrate, A solid piece of Glass or mirror could be shot that size, while marble since its a softer material would be cut into pieces to fit the scale of the job. We could etch the size of a football stadium, but the file would need to be cut up many times around the 6X10 parameter. And a file that large would never be opened in even the computer we plan on building with software to boot. Our largest job so far is 40FT long by 10FT high which is the reason for this thread. Registration is done through a grid shot onto the table through another vector program. The table is then datumed to .025 by .025, and then homed. The only limitation is if we shoot marble, there are different paramenter and lense focal points intensities and speed that are switched to that subsurface, so we have to keep the material shot during a run the same, cant mix glass with marble, and granite as each has there own properties and thicknesses. Though through our registration process, we can set many different home starting points and fire the laser to shoot the whole 6X10 ft, which may consist of backspalshes, photography, and fineart in all different sizes.

cyphyr - You hit it on the dot, I will see if I can do a combination of all the above to achieve a working file of that size otherwise chunks is the only way. Boxx or the mentioned Tyan computers will be a next step, but its looking more towards a software limitation right now that needs to be figured out. There was lots of good info. here to make that call.
Its too bad though the company we bought the laser from did not create a program like photoshop to run such big files...

lots - Yeah, lots we do. Quite literally have disected everything down to a science, months of testing to get the optimum results for each subsurface being ran, and workarounds for the computers we have to speed up work flow and still keep quality. The data is the main reason why this thread was started, by the end of all this, I will see how big of a file size I will be able to work with, bigger jobs then the one descibed above is a whole nother ball game, but there is always a way and an answer to a question or workaround to get the desired reults. Also, yes, if stitching 4 files together to work from is the only way, then that is what will need to be done, the things that makes this machine so great is the details that can be shot in such a large piece. Ide hate to limit it because hardware or software cannot run it. There is a workaround though and most likelly there will have to be a particular process to be able to get the desired results. That is what I and the team will have to figure out next. Thanks for your insight.


P.S. Thanks for the help and info. from those that added rather then detracted from the thread. It all was very helpful, in part, I didnt think I had to explain the machine and reasons of such when the main question was hugh file size 200DPI - what can run it and how much.. But I guess unless you get all the info. first, its hard to answer the whys and hows..

and to SZOS your comments of NASA and impossibilities made for a great laugh for everyone at work! :applause: We are waiting for your next insightful thoughts and comments on how things work. Im sure they will be highly intellectual and benefit this thread greatly~:thumbsup:

06-17-2007, 10:51 PM
I fully understand that you have to use a dpi in the 100's for it to work, as the laser dots are generally of a set size, the etched dpi would be more accurately described as a density; so if its not dense enough, it will be faint.

I know how laser etching works but I dont know the exact system youre using as you havent said. However I find it impossible to believe that you cannot de-couple the input file dpi from the output etched dpi. I would genuinely bet a substantial sum of money on it.

But still, if this is truly the case then I would look at getting some better control software which can interpolate, rather than trying to match the unreasonable source file requirements. Something as straightforward as doubling up the output to give 2 etched dots for every pixel instead of 1 would bring your requirements down to a level where it can be done.

06-18-2007, 12:42 AM
as lots and I mentioned, dont forget those fast hard-drives, preferably solid-state if-not then 15k drives.

Those are just as important as any other requirement.

06-18-2007, 02:55 AM
I find its good to know the hows and whys of the problem. Sometimes people think they need something more than what they really need. It helps put perspective on things.

Just to mention price for a minute:

Getting a new computer will probably cost a bit, especially if you need the hard drive system to function at the greatest bandwidth possible. Which it sounds like you might. It probably breaks your $8000 budget. Just guessing here.. I havn't actually priced out the spec. Instead, you may want to consider expanding upon your current system. It should be cheaper to do. I don't know if you posted the specs of the current computer, I dont recall anyway :P. If the current system is good enough, though, simply speeding up the disk I/O on your scratch disks, could help your problem. Then when the next system purchase comes up, you'll know where to invest the bulk of your money.

FWIW, solid state drives are alot of money. If you did manage to go that route, you're looking at several thousand for just the disk alone (depending on size). You've still got other requirements after that. Good RAID controller, for example.

06-18-2007, 03:57 AM
Ill be talking to boxx first thing on monday, and see whats the best route to go after that. I have more then enough info. here to figure out what we want to do. Thats a good suggestion regarding optomizing the machine I already have, but that I think is going to another designer and the new build is going specifically for the bigger jobs. As soon as we know the route we are going to take, and the build with software and hardware, Ill be throwing the specs along with any info. to help out those looking for a comp that runs great under these specific conditions.. Ill be looking at the bigger and faster harddrives as well.

to answer your question imationastion - you are correct in your first statement regarding DPI, the idea behind this all is to not see the dots, same goes for why giclee prints are so expensive. Trying to get the best quality out of the best image to print perfectly onto a surface, in this case, etched into a subsurface with specific perameters.

"I find it impossible to believe that you cannot de-couple the input file dpi from the output etched dpi." - Nothings impossible, everyone knows that. by de-couple, do you mean taking a 72DPI photoshop image and bringing it into the laser program and making it 200DPI. Sure, if that is what you mean, as the end result to the laser spitts out 1bit images as stated before for the laser to compute. The trouble about that though is simply like anything else. Since you know DPI now acts like density of the particular image based off the initial size, then you know if you are working with a 72DPI file and save it out, and transfer it to the laser and increase the DPI to 200, that your shooting 200DPI with inital 72DPI image from before, quality just would not be the same as saving 200DPI-etching 200DPI. This can go back to upsampling it. You lose quality, its states that in every book, even our lovelly cgtalk challenges asks us not to upsample our files for competition but to leave it at the resolution we created them at.
Open a 1200X800 image off this site somewhere with 300DPI, change image quality to 72DPI, then change it back to 300DPI, and see the quality change, in a sense, that is what you are talking about. Correct me if I am wrong and if you are talking about something else. In that case though to save on file size, at 150DPI to 200DPI would show less file quality loss. Hard trouble is the laser needs a minimum of 200DPI for the image to etch the best and be seen without pixelation. 300DPI is more prime, but too big file wise and takes longer to shoot. I dont know if I answered your money question, if not, please be more specific and I can tell you what weve done and the results.
Your pixelation idea is spott on, weve tried using less DPI and adding pixels to each pixel shot through photoshop by indexing color and increasing pixel size, but the results were mixed. Almost giving us a more blotchy detail. Though coming from that process we have worked on a coloring workflow completelly done through the laser and hand without artists having to paint into the stone manually, that is still being worked on to try to make it completelly efficient.


06-18-2007, 05:41 AM
A Quad-CPU (Core2Quad), 8GB RAM, 3 drives in RAID5 (750gb each), 8800Ultra in SLI.

That should run about 5k. 10k for a workstation is A LOT. :)

06-18-2007, 12:52 PM
A Quad-CPU (Core2Quad), 8GB RAM, 3 drives in RAID5 (750gb each), 8800Ultra in SLI.

That should run about 5k. 10k for a workstation is A LOT. :)
A pair of high end GPUs in SLI like that will not do anything for painting apps :P So thats a wasted chunk of cash (about 2k worth :P). Also, you don't need big hard drives, a set of 36GB 15krpm drives (8 or so?), and if you want absolute speed RAID 0 is the fastest (granted the most risky in terms of data loss). But its for scratch disk not storage :P

Basically though, the hardware is there to support this kind of image. Its the painting apps that are lacking 64bit versions.

06-18-2007, 02:36 PM
wow the stuff in this thread is crazy lol :p
I'm pretty this wont cut it for what you want, but maybe try project dogwaffle 4 (http://www.thebest3d.com/pdpro/index.html), there is a free demo, so i guess there is nothing to loose :)

06-23-2007, 11:04 AM
Too bad they discontinued Live Picture years ago. It sounds like a good solution for this type of work. It didn't require you to load your image into ram, it worked off a proxy type file keeping track of your edits. When you're done, you would just render out the final image at any resolution. Live Picture would take the original source file, combine it with your edits and then create a final output file. Sort of how rendering out video works. Live Picture even had a batch render control that allowed you to output to various formats and resolutions from the same file. Just make your output choices, hit the switch for an overnight render and hope it didn't crash. Back in the day, Live Picture had a decent following because ram was so expensive ($1200+ for 64MB). I remember working on 2 gig images with only 128megs of ram and a 100mhz processor. I was able to do my edits in real time because the proxy image a very small file size. It's like working on a 72 dpi image and then applying all of your edits to the master image that could be sitting on another drive. Imagine what you could do with todays hardware!

Live Picture ran on Macs and I bet you could find a copy and try it out on PowerPC with OS 9. It might be a little unstable, but if you optimized the system it might work. It would be funny if you solved your problem using old software and cheap outdated hardware.

06-23-2007, 04:41 PM
Live Picture was a very cool app and its too bad it got dead-ended

08-06-2007, 04:33 AM
well that last program sounded cool. And yes it seems terrible that is not around now. but what is going on with this thread?


08-06-2007, 09:23 PM
First of all I'm assuming that SEL is talking about a CO2 laser rig and that he's working with photos as these issues generally don't apply to line art.

Working in anything other than grayscale is useless with laser etching equipment and the resolution that you etch/cut at should be higher than the resolution that you edit your images at because you are dithering down to 1bit when you etch, using the spacing of dots to represent the various levels of grey kinda like a newspaper. It takes more than 1 dot to represent a single grayscale pixel. For granite murals etching at more than 200 dpi is going to wash out your detail. I'll usually edit at 72-100dpi for granite. For marble 600 dpi is the absolute max that I have been able to etch without washing out the image. I'll edit at 200-400 dpi for that depending on the size of the final product, larger mural=lower dpi because a larger mural will typically be viewed from further away than a smaller one. My machine etches at up to 1200dpi as well but I've never found a material that was useful at that resolution, even anodized aluminum loses detail at that rez.

I've been doing laser etching work for about 5 years now. I've done several pieces at the sizes that SEL is talking about (largest was 20x20 foot) and my hardware is an aging Athlon XP2800 with 1GB of RAM running Win2k. Data/swap drives are striped (2 drives, ata100) and I mostly use Corel 11 and have yet to get a job that couldn't be done using this crappy old system.

SEL, you might want to do some more testing to determine what resolutions work best for the materials at hand. Working at those super high resolutions is unnecessary. Are you using Photograv (or something similar) to do your 1bit conversion or are you printing right out of Photoshop?

I would think that a machine with 4GB of RAM should be fine. You're doing 2d work so a quadrofx card won't offer that much of a performance increase. As far as I know there aren't any laser systems that offer drivers for Mac or for 64 bit Windows so unless you're moving your file to another machine to control the laser you can probably rule those out.

What kind of laser system are you using, Vytek?

Good luck with your hardware quest.

08-07-2007, 01:57 AM
It's funny how no one asked question that begs to be asked: who in the hell wants full color marble etching in that size?! (or in any size for that matter) I can only imagine few russian tycoons, several columbian druglords or some of the G-Unit members... :)

Well, I'm sure you got customers, but it just sounds like final product is something extremely kitschy and cheesy (no offense meant at your drawing / design skills, it's the technology for final output that tickles my imagination).

Could you please link photos of some finished pieces (preferably in color)?

As for hardware - just one tip (already mentioned) - you absolutely don't need Quadro graphic card for work in Photoshop. You'll get same performance in Photoshop with Quadro FX 5500 and onboard graphics on cheapest motherboard on market. If you won't do anything but Photoshop work on that rig buy cheapest graphic card (people who put Quadro in your current rig tricked you, unless you're working with some 3D apps and massive amounts of polygons).

08-07-2007, 02:44 AM
Can't speak for SEL but I do very good business with granite murals sold to doctors, lawyers, architects and owners of upscale homes. When you realize that this type of work goes for $150-250 a square foot and that a 10 foot square mural can go for $15k plus fees for the photoshop work you'll realize that there is some serious money to be made here. I don't do color but there are some examples of color stuff here:


08-07-2007, 03:21 AM
yeah splitpoint is correct, there are lots of uses for granite and marble laser etching along with other materials. As an architecture student I can see a use for it. I mean something like a building lobby, offices, upscale homes, and various other uses.

08-07-2007, 12:12 PM
Like I said, I'm sure there are many well paying customers, and I'm sure it's a business which makes tons of $$$, but that link Splitpoint posted just proves my thoughts - it's EXTREMELY kitschy and cheesy.

I already knew how detailed marble etchings look (also very kitschy to me), but the color stuff...OMG

08-07-2007, 12:50 PM
So what happened? What did you end up going with? That is pretty amazing stuff. Honestly, It sounds like your company may want to invest in some custom programming (hire some programmers). I'd say linux is the way to go. Gimp is a pretty decent program, but you wiil want something that break the image apart like you want and all that.

It's like opening a 20 Million poly Z Brush mesh in Lightwave or Maya: CRASH AND BURN!

Good luck to you!

08-07-2007, 03:04 PM
Like I said, I'm sure there are many well paying customers, and I'm sure it's a business which makes tons of $$$, but that link Splitpoint posted just proves my thoughts - it's EXTREMELY kitschy and cheesy.

I already knew how detailed marble etchings look (also very kitschy to me), but the color stuff...OMG

Yeah to tell the truth I have to agree with you on that link. I didnt check it out until after I posted. I didnt see anything there that wasnt really cheesy. I was thinking more along the lines of a light pattern or design. the stuff there was too strong, from an architectural since I would question the astetics of a giant picture of a dog or something etched in granite.

08-07-2007, 04:36 PM
I could see why you'd think that. Typically alot of my work involves a business name or corporate logo, for an entryway or rwall behind receptionist. I also get some glasswork for restaraunts and memorials.

08-07-2007, 04:57 PM
yeah thats more along the lines of how I could see it used and thats what I figured most would do with it.

08-07-2007, 10:29 PM
SPLITPOINT - thanks for posting your real-life experience with the subject matter. It just validated what I posted originaly in this thread, yet I was quickly blasted by the original poster, who obviously has no clue what he is talking about, and was looking for a PC that was absurdly powerful even when it wouldn't really be necissary. Good to know that there are people out there, like you, that actually know how to use this equipement, too bad the original poster is not one of them obviously.

08-08-2007, 02:54 AM
I figured Mr. Szos would eventually chime in. Good to know the guys still doesnt have a life! :thumbsup:

To Splitpoint, Its nice hearing from someone else who uses im sure a simliar laser. We Have the Vytek L-Star Laser in The Wearhouse, 1-5 in the US at its size at 6FOOT Tall by 10FT wide. I dont know if you are one of the other companies that has one but I know it is very costly. Majority of our competion has the one hafe its size which they are happy with, trouble is if your doing huge glass enclosures or mirrors etc, they would need to be cut down to size to fit the laser where the bigger ones will not. Majority of our work comes from the businesses you described, architects, interior designers, very wealthy people. To chime in on your computer specs, thats nice you have been able to shoot 20X20 ft murals im assuming, the thing is with that, im sure your original file is cutup many times to shoot that size on 12X12granite on your laser. With that said also, we convert to greyscale once the overall image is approved by client, the color image in necessary to get an exact replica when in the painting process. Used as reference. The image though is then sized to specs, and then worked on in greyscale. We do a form of Matte Painting with our murals mostly because our clients looked for original art of sometihng they imagined. You are right in a sense that we rarelly use 600DPI or higher, it just isnt necessary, but to shoot something on granite less then 100DPI with a laser just wont give you the quality needed on a big job. If you are fine with the detail looking from a farther distance, it is fine, but we have had too many requests to where they want it extremelly detailed upclose on the big stuff. Its a catch 22, if you want the 20k job, you find a way to do a 20FT mural with extreme detail in every spot. That was the reason for this thread was to find a solution to that problem where we ran into a tight deadline to create a 16FT by 8FT Mural of a Matte Painting done on computer that every inch uplcose would have detail, it sounds like you would not go to this extreme on your end, and thus why your computer is capable of doing the work. I wouldnt have made a thread about this unless I needed help on specs. 200DPI was the specs we needed to follow, and everyinch of the matte painting needed to be perfect on granite before paint. Guys spending that much money tend to want things a certain way or find someone else who can. We had someone build a machine based off the specs from this thread that fit the job perfect. Money paid, job finished, onto the next.

End of story, thanks to those again who contributed, those that are interested in the process, look online to find resources on laser etching. There are many great businesses that do amazing work that you can get to know the process better, every laser is different, but an overal grasp of the capablities are endless.


08-08-2007, 03:35 PM
SEL, do you happen to have a pic of the final product? Or even the product in some of the stages as it's being created?

I'd very much like to see the process and detail of the work involved.

08-08-2007, 04:25 PM
Yeah, I'd actually like to see some samples too. The whole process seems interesting :)

08-08-2007, 05:04 PM
yeah same here, im assuming its similar to a laser cutter, but the ones we have at school can only do like 30x14" piece and it cant be used on anything but some woods, museum board and acrylic up to 1/8th in thick. ill be interesting to see how you can do something so large and dense as granite or marble.

08-08-2007, 05:47 PM
The hardware is the same as a laser cutter, it uses different driver software to rasterize the image. They're available in various wattages and table sizes, Universal and Epilog manufacture them with the small size tables and Vytek and Camtek make the machines with the larger table sizes. There are other manufacturers but those are the big ones. The largest table size that I've personally seen was 6x8 foot but I know that larger tables exist. Basically it's a giant plotter with a mirror and lense attached rather than a pen. The beam is directed to the surface of the material using mirrors and a lense.

Interestingly, you don't need high wattage to etch stone. My rig is a 35watt Epilog and the way it works with granite is that it heats up the surface and causes the stone to crack/explode which turns colored granites white. Black marble has hydrocarbons in it so the laser heats up the surface which cracks the surface and bleaches out the hydrocarbons turning it white where the laser touchs the stone. Anodized aluminum is bleached using the same process. Wood is just vaporized by the heat of the laser. Using th e laser on glass fractures the surface which frosts the glass. Steel requires a ceramic paint that is fused/bonded to the surface using heat from the laser. Since I use a small table format I tile my large images by bringing the full size image into Corel, setting the paper size to the size of my table and then move the image numerically the distance of my table size. Sounds more complicated than it is.

Using the CO2 laser as a cutter works with thin plastics and wood but is impractical for most other materials. If you need to cut metal of any thickness you'd need to look at a YAG laser or water cutter.

08-08-2007, 06:36 PM
For you guys asking for examples....
First is a mural that I did for a guy on granite, second image is etched glass that was hand painted to colorize, sorry for the washed out color but it was really difficult to photograph that window. Third was another mural that I did for the first guy.

08-08-2007, 07:12 PM
A couple of more things and then I'll shutup.

I'm working full rez all the way through the process, the image is only tiled when it goes to the "printer." It also seems to me that by working at the same dpi on your image as the final output you are probably reducing the quality of your output, especially on Granite. When you go to print the software dithers the output file to 1bit so that the laser controller can understand it, if you're dithering at the same dpi as the source image then you're image is losing resolution at conversion time. You might want to experiment a bit more with the settings before writing this off. If you're editing at 200dpi you should be printing at an even higher resolution in theory but the physics is such that past about 3-400dpi on Granite you'll wash out the image because the width of the laser beam is as large or larger than the granite crystals.

Hope you get your hardware issue figured out.

08-09-2007, 06:13 AM
Ill throw up our website here sometime tommarrow so everyone can have a better idea of the process etc. of what we do. Web guy is finishing things up copy wise and gallery as there was a major change with the site.

Laser Specs are 120Watt laser 6X10Foot bed. Up to 1200DPI - L-Star Vytek Laser. Our processes are many, depending on what the client wants. Our biggest seller are deap stone rugs which are pretty amazing in their own right as we can create any design we want with a bevel for flooring and entrances. Our white marble is pretty amazing too as that material can use the process also creating amazing looks. The other stuff minus wood break the top surface of the stone (ETCH) creating your image. Wood is amazing in itsself, we have not ran enough tests with it, but once that part of the business opens up, it will be groundbreaking scale wise. Ill put that up in the gallery for you guys to check out also. Its like taking any image you have 6X10FT high and creating different depths to the wood using a vector art. Ill be posting more shortly.

Its good hearing from splitpoint as he understand what goes into this stuff, while some of our own processes may be different, the overall ideas are the same, numbers do fluctuate depending on material and your machine you use though.


08-13-2007, 11:57 PM
ok so I don't understand this one. If the end piece is going to be one bit then why not as other people have said work in grayscale? Yes of course there must be workarounds and all that but it seems to me you are going the wrong way round. Build on the process of others. Learn from people and then see what you need to change so that the process works for you.


CGTalk Moderation
08-13-2007, 11:57 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.