PDA

View Full Version : Confused about HDR making: f-stop or shutter speed?


lehthanis
11-18-2004, 05:09 AM
So I'm going tomorrow to shoot some light probes with my nice new mcmaster-carr 3" chrome steel ball right? well...I've read a tutorial on here in that other thread (http://www.hello-napalm.com/tuts/tut_hdri.html) that says you shoot a range of shutter speeds (which refers to them as f-stops) whereas my knowledge of photography says f-stop is aperture, different from shutter speed...and the debevec tutorials say f-stop which is also what HDRshop is set to use...

Which gives you the best results? f-stop sequences or shutter speed sequences?

Whichever one you go with...what setting do you use for the other option? whatever the recommended setting is for the first correct shot?

Thanks in advance :D

Andrew W
11-18-2004, 09:31 AM
You should shoot with different shutter speeds as different f-stops have varying depth of field so areas could go in and out of focus which will lead to artifacts in your HDR image.

The method the Debevec team use is to place place combinations of neutral density filters in front of the lens to attenuate exposure.

Hope that helps,

A

stew
11-18-2004, 09:47 AM
If my memory serves right, f-stop or aperture is often being used as unit for exposure differences. At least in German among the photographers I know, you make an image "one aperture brighter", even when you're doubling the exposure time and leave the rest as it is.

lehthanis
11-18-2004, 02:33 PM
So we're divided...now, knowing about aperture and how it affects DOF based on shutter speed, I'm inclined to believe that its the shutter speed that should be altered...and I'm guessing that it doesn't matter as long as a proper curve is calibrated for the camera based on the shots, right?

Andrew W
11-18-2004, 03:59 PM
So we're divided...now, knowing about aperture and how it affects DOF based on shutter speed, I'm inclined to believe that its the shutter speed that should be altered...and I'm guessing that it doesn't matter as long as a proper curve is calibrated for the camera based on the shots, right?
Each additional f-stop of exposure doubles the amount of light hitting the film plane. e.g. f5.6 permits twice as much light as f8 to hit the film. Similarly (and more obviously) 125th of a second exposes twice as much light as an exposure of 250th of a second. If you want to use neutral density filters these are measured in "stops" of exposure also.

Therefore in terms of the relationship of one exposure to the other either method is valid. I favour altering shutter speed because of the depth of field issue I mentioned before. The fact that HDRshop mentions f-stops doesn't matter as long a you realise how each f-stop is related to another and apply that to your shutter speed selection.

Andrew

lehthanis
11-18-2004, 06:19 PM
Good call Andrew, thanx for the clarification. I'll use shutter speed.

CapnPanic
11-18-2004, 07:01 PM
Sorry if it was confusing. I was using f-stop as an exposure measurement of light rather then an aperture measurement. I do recommend changing the shutter speed and not the aperture for the reasons mentioned above :)

If there are any other bits that you found confusing, let me know so i can reword the tut and fix em :)

lehthanis
11-18-2004, 08:27 PM
It wasn't necessarily confusing, it was just that I had always known HDRs as being done with f-stops...if you clarify that, it should be alright.

You might want to mention the neutral density filters that debevec used.

rendermaniac
11-18-2004, 11:06 PM
There is quite a good explanation here http://www.uscoles.com/fstop.htm . Andrew is dead right though - shutter speed all the way. Plus you have a larger range of shutter speeds than apertures. Doubling the shutter speed or opening the paerture on stop increase the exposure by a stop. I think this may be why quite a few cameras companies prefer to refer to change of exposure in EV's (exposure values - it's just a different name for a stop) to reduce confusion.

This is a bit like the term HDRI which is an image with a larger range of exposures than your "normal" 8 bit image. Most people mean a map for image based lighting that happens to also be high dynamic range.

Also use as slow a film as you can find if you are shooting in sunlight as capturing the highlights is the difficult part. I have used ISO 32 film before, but even this wasn't quite slow enough. The shadows take care of themselves - ironically as you have no real limit on how long your exposures are! (would be interesting to see a night HDR image).

Obviously things like people, cars, and strong windws (clouds, trees rippling, tripod wobbling) are best avoided if possible.

Paul Debevec used neutral density filters to reduce the exposure by 4 stops, but these filters are very hard to find (Kodak Wratten filters).

Andrew - I read on cgnetworks.com about you using image based lighting on Troy, but it said you only used 5 exposures - is that right? Even shooting at 3 EV intervals and processing as RAW files did this have enough detail to capture all the highlight detail? Or was it just yet another misleading vfx article? ;)

Simon

lehthanis
11-18-2004, 11:10 PM
Ran into a problem taking my HDR tests today...

My wifes camera has two lenses...a 28-80mm, and a 75-300mm, both decent lenses...but There is no way I can fill the frame with my 2" ball and be in focus. I can't get close enough and focus on it to get the shot full frame...I can get it about as big as the crosshairs in the lens (3/4th)...whats the best kind of lens to use for a 2" ball? or do I just need a bigger ball?

Also, what do ya'll use for a stand for the ball...I used a cardboard tube taped to a maglight ad set the ball on the tube, makeshift...gonna cut a broomstick and pt a point on one end to stick into the ground and some of that poster tack on the other end to stick the ball to. Any better ideas?

rebo
11-19-2004, 12:02 AM
Get a bigger ball. 4" to a maximum of 10" imo.

Remeber while a large ball will enable you to focus better, it has a larger blind spot behind it as the tangent angle is increased.

lehthanis
11-19-2004, 01:35 AM
Wow, very helpful! Thank you guys a lot. Now...until I get a larger ball, am I ok with not filling the frame? I know my registration will be a bit more work, but as long as I scan the negatives large enough I'll still be able to isolate the ball enough to get a good scene right?

You mentioned EV's...and this camera has three shot exposure bracketing...if all I want is basic lighting information that I'm going to diffuse anyway, and one good hi resolution background, will three exposures 1 EV apart each be enough for a simpler scene?

CapnPanic
11-19-2004, 05:53 AM
i recommend going to a garden store and picking up one of those mirror ball lawn ornaments. They are not very optically true, but work pretty well nonetheless.

you can get away with fewer exposures, like the 3 you mentioned, taken one stop apart, but just be aware that the less overall exposure you cover the less dynamic range your final hdr image will have. although i think you will get better results if you can make it bracket 2 or maybe 3 stops apart, so your 3 exposures cover more dynamic 'ground'

on a side note, if you use a digital camera with the option to save the photos in raw format, you can take fewer exposures (say 2 or 3 stops apart) and as long as you do your image processing in photoshop CS, where you can keep everything 16-bit, you will pass the extra info into hdrshop and can save some time. this would also be handy if you were on set and could only squeeze in a few shots.

Andrew W
11-19-2004, 09:30 AM
Andrew - I read on cgnetworks.com about you using image based lighting on Troy, but it said you only used 5 exposures - is that right? Even shooting at 3 EV intervals and processing as RAW files did this have enough detail to capture all the highlight detail? Or was it just yet another misleading vfx article? ;)

Simon
No, that article's correct but generally we used a CG light to act as the sun because our HDR images weren't HDR enough to capture the intensity of the sun itself. I favour this approach as relying purely on HDRI for a lighting source doesn't give the artistic control that I want.

Andrew

rendermaniac
11-19-2004, 09:43 AM
That makes sense. I guess if you have the sun "baked in" to your map then it does limit what you can do with it. I guess something like HDRgen gives you a good starting point though.

The one case we've seen that it helps having this detail is perfect mirror reflections as specular just looks like dirt. Then again using a 2D cheat is probably a better solution.

Simon

lehthanis
11-19-2004, 02:49 PM
heh, the camera has bracketing of +-2 EV's maximum. though you can combine that with exposure compensation to a maximum of +-3. I believe though that exposure compensation will shift all three shots in only one direction not both. Its a Nikon N75, if anyone knows.

As an aside...my wife just won a Nikkor 50mm f/1.4 1:1.4 lens last night for 50 bucks on ebay, yay for ebay!

MooseDog
11-19-2004, 04:29 PM
chk this out for some impressive balls:D


http://www.outdoordecor.com/cgi-local/SoftCart.exe/online-store/scstore/p-RI712-S.html?L+scstore+bxmb3541Google-Decor-GazingGlobesff670467+1058693988

Andrew W
11-19-2004, 05:21 PM
Paul Debevec used neutral density filters to reduce the exposure by 4 stops, but these filters are very hard to find (Kodak Wratten filters).
I've been thinking about this. I reckon you could buy a few sheets of ND lighting filter (transparent sheets of plastic essentially, used for "stopping down" lights) from Lee or a similar manufacturer and cut it into smaller squares that you could layer up in front of the lens in combination. They are not optically perfect but they're almost certainly good enough, and pretty cost effective too.

Andrew

rendermaniac
11-19-2004, 05:49 PM
I've been thinking about this. I reckon you could buy a few sheets of ND lighting filter (transparent sheets of plastic essentially, used for "stopping down" lights) from Lee or a similar manufacturer and cut it into smaller squares that you could layer up in front of the lens in combination. They are not optically perfect but they're almost certainly good enough, and pretty cost effective too.

Andrew
Tried that ;) The problem is flare - you get an awful amount of it and the images come out unuseable unfortunately. It's because the light bounces between the layers. Very annoying!

Simon

lehthanis
11-19-2004, 08:54 PM
chk this out for some impressive balls:D


http://www.outdoordecor.com/cgi-local/SoftCart.exe/online-store/scstore/p-RI712-S.html?L+scstore+bxmb3541Google-Decor-GazingGlobesff670467+1058693988
I saw that site, thinking about the 6"

lehthanis
11-21-2004, 03:18 AM
I just bought a Nikon E4200 Coolpix camera...its only 4 megapixel, and it doesn't have raw format, but it does have a decent zoom on it...I'm able to fill the frame with the 2" ball, but its not 100% accurate. Plus I need to work on the actual dynamic range portion and registration.

Andrew W
11-22-2004, 09:34 AM
Tried that ;) The problem is flare - you get an awful amount of it and the images come out unuseable unfortunately. It's because the light bounces between the layers. Very annoying!

Simon
I was thinking of building some kind of matte box (you can buy cheapy plastic ones from any photo supplier) and then sandwiching the filters in that. I think for ease of use I'd suggest pre-building the differing layers of filter so you can just swap them in and out at will. I've used plastic filters like these before on movie cameras with excellent results. I think it's a matter of how you mount them not an inherent problem with the filter itself.

A

rendermaniac
11-25-2004, 08:48 PM
Ah I was still thinking fisheyes! Yeah with a long lense and a light probe that would probably work. I guess getting a darker mirrorball might work, but getting it even would be hard.

Simon

cgJD
11-26-2004, 01:21 PM
The problem I have found with mirrorballs - certainly the solid ball bearing ones (mine is about 10cm in diameter) is that they scratch very easily - tiny tiny microscratches that cannot be seen easily by the naked eye but distort the highlights on the ball in bright light. Be very careful when handling the mirror balls.

I always use the smallest apeture (largest F-stop) to give me the largest number of pictures before the camera whites-out. Ideally the pictures should go from black to white to ensure that you have captured the largest dynamic range you can. If the shortest exposure times still have highlights / detail then there is still more dynamic range to be captured.

With my camera I can take about 45 different exposure levels - by taking photos at all or most levels you can remove any pictures that suffered from natural light level changes (because of clouds etc). You can see this problem occurring when HDRshop puts your pictures in the wrong order (it automatically calculates the overal brightness of the images and sorts them accordingly).

rendermaniac
11-26-2004, 11:00 PM
You may want to open up a couple of stops rather than choosing the smallest if you can as you can loose quality through diffraction.

Simon

lehthanis
12-02-2004, 04:53 PM
chk this out for some impressive balls:D


http://www.outdoordecor.com/cgi-local/SoftCart.exe/online-store/scstore/p-RI712-S.html?L+scstore+bxmb3541Google-Decor-GazingGlobesff670467+1058693988
Caveat Emptor - Buyer Beware

My 6 inch version came in today and it is SERIOUSLY flawed...it has long circumference bands of warping in several places. I would probably say its unuseable for anything even vaguely realistic. I'm goign to have to just rely on the 2 inch ball. since I can't afford a larger machined perfect one. (3 inches is 99 bucks!!!)

CGTalk Moderation
01-19-2006, 11:00 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.