True DOF

[RichMan]

Hi


Thanks for the Tut Man :)

:applause:

[FusionDG]

mdme_sadie,

thanks for the toot. Looking FWD to getting R8. I was disappointed that R7 handled DOF so poorly. It appears that it can be done in R8 with a little finagling, but it’s too bad that it requires these extra steps to get DOF to work.

I am a previous user of ArtLantis, and I don’t recall that program’s DOF being as hard to use or problematic as C4D’s DOF. But that program was short of some professional quality, so maybe I just never picked up on it's shortcoiming's in the DOF dept.

BTW, you are one the best narrators of toots that I have followed through. :thumbsup:
Everything is thoroughly explained and worded in layman’s terms. If you wrote a book on C4D I would buy it! Or maybe you already have one out?

Nice work! :beer:

[Per-Anders]

Thanks you're welcome, i've updated the tutorial slightly to give a smoother result (and the c4d files attached) enjoy.

[flingster]

Once again....you've come up on top bud.:applause:
Thanks for taking the TIME....and i mean time as with all these things they eat time most of us don't have.
So thanks again for sharing your efforts with the rest of this forum.:thumbsup:

[michaeli]

Very useful tut and very easy to understand. Thanks, mdme_sadie. :beer: :thumbsup:

[Per-Anders]

You're all welcome, and now I've just added another tutorial for controlling exposure on a camera by camera basis. This is an Xpresso tutorial. It's not the quickest way to go about things, but it does make things easier if you've got a lot of cameras in your scene, and it's a long animation.

http://www.peranders.com/c4d8/

Enjoy

[flingster]

this is probably dumb....but Xpresso is for what exactly?
Are we talking programming knowledge of drag n drop?
(i suppose its r8 specific?)
:eek:

[Per-Anders]

Yup it's V8 only.

Xpresso is a node based expression system. Basically it means you don't need to know how to program in order to get certain effects. If you look at the tutorial you'll see what's going on there, it can look mabe a little daunting at first, but once you get to know it it's very much simpler and easier than using Coffee. In it's current incarnation it's not as powerful as coffee, and it's not meant to be a replacement, but for animation it's very useful to have one attribute control attributes of other things (at it's simplest level for instance the rotation of a lens barrel controls the z movement of the lens), and to not have to program that, but simply drag a line from the Barrel rotation to the lens z position.

Node based expressions are one of the main strengths of packages such as Maya, Softimage XSI and Houdini. Here because C4D was never designed around a node based system from the off it's had to be implemented as an expression tag, rather tha something inherent and constantly behind the scenes and behind each and every move (as in the aforesaid apps). This is fine for most people, but at some point I would presume that Xpresso will come of age and have total control of all aspects of the package like those other apps.

[flingster]

OK....got it....thanks matey.
sounds very useful.....and likely to get more and more powerful over time. I don't think i'll be anywhere near this sort of stage for awhile yet....but when I get my copy of R8 i'll certainly give myself that baptism of fire, and get my feet wet.:wip:

You can just see some serious results coming out of the developments made in R8.....very exciting i think.

[Pate]

Thanks mdme_sadie for these superb tutorials!

Your new tutorial convinced me that I need to start playing with Xpresso when I get my R8.. Previously I had thought that I can continue using Coffee and not learn Xpresso, but now you made me see the light! :)

Thanks!

Pate

[H. Ikeda]

Hi,
Nice tutorials! Easy to read and inspiring (those are rather tips, i think). Thanks.
I have questions for each by one, though.
DOF: This may be a dumb question, but I couldn't understand the 2nd and 3rd figures therein. No, I understand what you mean by the figures. Rather those figures look strange to me. Z-buffer values are almost the same for a cube and floor location at which the cube stands? If it's true, why blur strengths are different for the both?
CE: Could we use Stage object's Camera parameter directly without User Data 'Camera'?

see you:cool:

[Per-Anders]

Z-Buffer is exactly the same for the floor and where the cube stands, that's why it's wrong. if you look at the examples you'll see that the reflectino of the cube in focus is out of focus as a result, whic just isn't real, also the cube that's out of focus in the background has a reflection that's pin sharp. that's just not what depth of field is like in real life, so by moving the camera around in a controlled fashion while keeping it targeted on a point you get real depth of field as you would out of a real camera, where reflections are accurate etc... and transparency has good dof too. the reason why it's really wrong in the z-buffer image is because z-buffers can only take the surface point as the bit to blur. they do a blur in one pass, as yet there's no post effect blur that i've come accross that makes seperate z-buffer channels for surface, reflection and transparency and uses them to blur the diffuse layer, the reflection layer and the transparency layer in turn (which would give you true DOF too).

basically the surface bluyr is correct in z-buffer type blurring systems, but anything reflected in the surface isnt. if you look at the reflection in the foreground of the true dof example you will see that the floor is blurred, but the reflecction in it isnt.

as far as using the stage objects camera setting, you could but the network would bloom, out massively, what you would need to do for that is make compare nodes that would compare each and every camera in the scene to the camera in the stage object, and then you owuld have to compare again to work out which was the correct camera to take the values from... it would end up even messier than it currently is (which isn't ideal, but works). also the condition node only takes in a numerical value as it's switch... you could do it, but it would be less than intuitive to extend the network to add more cameras.

i hope that this clears some things up :)

[amapimaster]

FusionDG - I have worked in artlantis a bit. Damn that small preview window!

[H. Ikeda]

Thanks a lot, mdme_sadie.

true DOF: I now understand! Z-buffer value is in proportion to Y coordinates of pixel's XY-plane (except the nearest cube) in the figures, so the same blur strength appears along the same Y level, regardless of surface or reflection. Not the case for true DOF, as you noticed.

CameraExposure: I see. That's the reason why we need User Data.

Thanks again:cool:

[H. Ikeda]

flingster: Here is an example of half-true(?) DOF in R7. The blur isn't plain in R7, but you could see sharp reflection for the second farthest cube and not sharp for the farthest. This uses less sophisticated method than mdme_sadie's ('half-true' means not so cool;)). This will work also in R8, as it is. That is, without expression, by using multi-pass rendering, the reflection layer has been blurred separately (mdme_sadie may suggest?). If necessary, I'll explain details here or in R7 tips thread.

[flingster]

Hey H. Ikeda thanks for taking the time for this.
Any chance you can give any more details.

As i mentioned i intended to have a bit of dabble later when 1) i get r8. Its really cool that someone took the time to talk about DOF cause up until this point all i've really heard was cinema was rubbish at it! So it really makes a change to see people sticking with it....albeit with r8 which i understand has improved this part dramitically. I know you might not want to but H. Ikeda is there any chance i can take a look at that example pictures cinema file? It would allow me to check out what you have done setup wise....if not, thats cool...but i always find it easier to learn this way...it then allows you to tweak/experiment/play and mess up etc.

Once again thanks to H. Ikeda & mdme_sadie for taking time to do stuff like this which improves everyones understanding of such things. I was thinking sometimes i wished maxon produced a reference book filled with examples of what can be done with each toolset....a sort of tutorial come example filled tool bible. Yes manual is good...and their tutorials give you a method skillfully glancing over the powerful parts of the tools. Just feel it would add to the tutorial which you sometimes just concentrate so much on what youre doing and do to do it that you often miss its capabilities.

[flingster]

by the way some of you have probably seen this before but lightwave has a plugin called x-dof which is pretty slick.

http://www.evasion3d.com/xdof_intro.html

But its worth visiting this site just to see some great examples of dof in the gallery. nice little 3d reference.

ps. i went on support and under development section it mentions the cinema briefly under platforms....does this mean they are planning cinema support....or what. Not that this is probably needed as r8 is supposed to be very slick also! dunno just a thought.

[H. Ikeda]

Thanks for your comments, flingster.
I've posted some explanations in
http://www.cgtalk.com/showthread.php?s=&postid=236756#post236756

Sorry I don't think a single c4d file will help you, because this includes several steps.
(But if necessary I'll post several files.)

cheers:)

[flingster]

ok thanks bud...i think i get what youre talking about now anyways.

[H. Ikeda]

mdme_sadie: I realized the problem of 1) in the 3rd figure is still not very clear to me. This may come from intersection of focal plain within object? This is peculiar to C4D 8? If so, in multi-pass method for R8, easy way of resolving this is only to set focal point outside objects in a scene?:curious:

cheers:bowdown:

[Per-Anders]

it's just a jump in the shape of the cube caused by the difference in how the dof handles foreground over background dof.

you can work around it by using the gradients to make it all a background blur (i.e. foreground blur distance set to 0, background blur set to where it would be, then using gradients to make it go from blurry near, to focussed at your focal point, to blurry again... i even made an xpresso network to do this using a couple of nulls, though it wasn't just a drap and drop affair, you had to set up the post effecct first, and then bring the post effect in, but it did enbale you to set a focal point with a null/sphere, and the blur points on the camera)... i might post it sometime.

post really is the fastest way to go, but my method gives you an even more accurate model of dof, and more importantly helps people to understand a bit of xpresso and it's power.

[H. Ikeda]

Thanks, I could get it. Further suggestion, too.
Originally posted by mdme_sadie
but my method gives you an even more accurate model of dof, and more importantly helps people to understand a bit of xpresso and it's power.
Everyone would agree with you. After all, multi-pass method might be 'half true.' I also think XPresso is a key tool in R8+. :)

[CGTalk Moderation]

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.