hello,
I am rendering with a vrayZdepth pass enabled.
while my image looks nice and smooth - the z pass is jagged and aliased.
Here is a the file:
http://proforma.preset.de/vRayBoard/20061122/rdg-aliased\_zdepth.zip
I don’t know why - but this is the first time this happens.
Am I missing some crucial checkbox?
regards,
Georg
yep, its a pain.
Apparantly its more correct without aliasing because when it has aliasing it adds tones of gray on all the adges…and tones of gray mean depth.
But…I couldnt get a nice result using non aliased image in photoshop to create a DOF effect, on the edges it was all messed up. What I did to make it aliased was to render the Zdepth at 12k wide and down size it to 5k image (which was the size of the image I needed to blur). Downsizing the image put aliasing back into it. It also worked very well with the Lensblur in photoshop.
this if of bo use if your doing animation, I was only doing one image.
update:
I now looked at older zdepth-passes and they look the same.
As it didn’t bother me in the past - I should not bother me now …
Georg
changing the image sampler to ‘Fixed’ - solved the problem for me.
zdepth isnt supposed to be antialiased since the blended pixel values at the object borders would introduce wrong depth information
combustion (and i guess other compositing tools, too) use the subpixel weight and coverate channels to come up with good results when depth blurring (hence those channels are suggested when doing post dof)
other than that there always is the trick with doubling output res and downsizing after the blur - if you can afford the render times and are able to render at those resolutions
btw: does post DOF still look good at 5k? i’d rather stick to distributed rendering with real dof (esp. since good depth blurs a’la lenscare tend to be slow, too)
subpixel weight and coverate channels
Are they available as passes in vray?
:lol:
???
After reading in the forum I see that this is a complex thing.
The answer could have been:
no they are not available, because of <bla>. so a common way of DOF is <bla> RPF <bla> <bla> combustion <bla>.
Fixed image sampler rules.
Don’t have combustion - so why caring for strange channels.
Georg
ive been asking and asking if someone could either do or point me to a tutorial on the sub-pixel or coverage workflow within combustion. I think i understand the concept behind it but not the workflow. just a screen cap of it in use would be all i would need really.
i m sorry for the sarcasm…
but if by passes u mean render elements u can easily find ur answer:
readme vray 1.5 file:
“Render elements
The 3dsmax Render Elements interface is now supported by V-Ray. The old G-Buffer settings have been removed.
VRayMtlSelect render element allows the user to extract a given material to a separate render element. This can be either a top-level V-Ray material, or a sub-material of VRayBlendMtl.
VRayWireColor render element allows the user to create masks with specified colors.
Color mapping per render element. Color mapping can be turned on or off on a per-element basis.
Render elements are fully supported in distributed rendering mode.”
Re: vrayZDepth alias
First, turn ON the “Enable Filtering” check box for the ZDepth render element; next, switch to Adaptive Subdivision sampler and check its “Object outline” and “Normal threshold” options. This will make it fine.
You need the “Enable Filtering” option in order for V-Ray to actually perform antialiasing on the z-depth channel.
You need the Adaptive Subdivision sampler since your RGB image is black and in that case the Adaptive QMC sampler takes just one sample per pixel. On the other hand, when you turn on the “Object outline” and “Normal threshold” options for the Adaptive subdivision sampler, it will take more samples at object boundaries even though the RGB channel is black.
Best regards,
Vlado
afaik it gets used automagically if present in the rpf file
quote from the combustion help:
Note: In Box Blur mode, the 3D Depth of Field effect uses only the Z (Z Depth) channel.In Gaussian Blur mode, you must also use either the Non-Clamped Color or Color channel. The Coverage, Transparency, and Sub-Pixel Weight channels, though not required in Gaussian mode, are suggested.
Most of the 3D Post effects that use the RPF channels Color, Transparency, and Sub-pixel weight do not require these channels to be present (they are not available for RLA files). If present, the channels are used to improve the quality of the effect.
@ vlado:
Thank you. This is even better than the fixed sampler
@ ene.xis:
sarcasm - no problem. It is hard in such a multi-lingual-english-community.
Georg
again, sorry bout that…
hope u sort it out…
mike.edel
“The Coverage, Transparency, and Sub-Pixel Weight channels, though not required in Gaussian mode, are suggested.”
for somethng thats not requiered but suggested it doesnt sound too automatic to me otherwise they would have stated plainly “subpixel and coverage channels are used automatically when using files that store those channels”
you seem to forget that it is discreet documentation you are talking about
in my eyes they sometimes aren’t too gifted when it comes to docs (the mental ray docs come to my mind… *shudder*)
alas: i did a quick test:
simple scene, consisting of a checkerboard ground and *tada* teapots
image 1 with zdepth, color, real color, coverage and subpixel weight (which seems to be empty but that might be my mistake)
image 2 just with zdepth, color and real color
both got a 3D Depth of Field Operator with identical Settings on top (used 15 gaussian blur to emphasize the difference)
as it seems, the gaussian blur doesnt work at all without the additional channels - there was no difference to the original image
box blur worked, but looked really ugly (on both inputs)
will eventually upload the images tomorrow - now i’m off to bed
mike
it worked ok and it was quick (about 2-3 mins), I was up against it with a deadline, which was the main reson for doing it in post..
vlado, cheers for the post..I wil try that next time
im an AE user who is always curious about combustion and the need to upgrade. lenscare does good DOF but it doesnt use the coverage channel. im just curious if lenscare using an AA filtered z-depth image for DOF is as good as combustions standard DOF utility using the coverage channel.
Lenscare is miles ahead of the rpf focus blur in combustion. You should render your zbuffer out in 2x rez like mentioned earlier in this thread (and then scale the layer down to fit the comp rez if you wish). This way it calculates the depth of the original res, and resamples it to fit in the comp. Twice as slow but very good quality. The layer with the zdepth footage can then be turned off if you are not going to use it in your comp for anything else.
I’m using VRay 1.47.03 + combustion 4 but it always been a problem to have any real info about coverage and sub-pix weight/mask in order to use it with combustion’s Motion blur or Z-Depth…
Actually I have 2 main problems :
- the Z Depth channel of the rpf file rendered by VRay is allways white.
- the ID Material selection made in combustion are really “dirty”, unprecises and unpredictables. Especially when a lot of objects are in front of the other in the scene.
If someone can help me…