Hi,
have you guys noticed that when “dont affect colors” is checked in Color mapping, then no matter what method and burn you choose (reinhard, linear, 0.35, 0.75 whatever)…the render always looks the same?
As if this turned off color mapping completely…
is this correct?
Thanks
Martin
yes, there was a thread about this, with “dont affect colors checked” all color mappings are ignored, except clamp i think
My understanding is that this is correct - the colour mapping is tone mapping work, which adjusts the full linear data of the file, and if you want a “pure” linear render you don’t want those changes baked in as you will lose (for example) highlight detail.
Anyway, I’m actually moving away from LWF rendering. I’m sick of looking at washed out renders that always need extra post. I just never get the contrast looking good in LWF without post, and creatively it’s challenging to establish materials and lights with a crappy looking image 
I’m still rendering 32bit, but with the colour mapping etc applied now.
/b
yes, that’s what it does, it doesn’t affect colours. vray will still do the correct sampling, knowing that you’re rendering, say, with corrected gamma, but won’t apply any colour mapping to the image. as an option is basically meaningful only if you render with linear multiply and gamma correction, and then save to a linear file format.
thanks for the explanation guys!!!
@simmsimaging:
Hi Brett…by saying you moved away from LWF do you mean you completely got rid of the LUT setup too? Or do you just omit checking "dont affect collors and gamma at 2.2 in color mapping tab?
Kinda woder what you mean by washed out result..i am getting images full of contrast now actually…can you post an example?
Best
Hey Martin -
I am still using the gamma setup in Max but have the image output set to 2.2, and in Vray “dont affect colours” is off. Now I can use Reinhard mapping and have the burn/gamma settings affect things. I typically reduce the gamma setting there to 1.8 or even less to increase the contrast in the image. At 2.2 I just can’t seem to get what I want.
I will try and post an example later.
have you tried using float values for the colors when you use them in materials?
this is especially with dark colors very important to use them
That’s something I only recently learned about for getting decent blacks, but it’s not just the blacks. It’s just hard to get nice mid-tone to 3/4 tone depth with LWF, at least for me.
Hi all,
after 4 years of 2.2 I am pretty sure that the LWF as we know it in VRAY is wrong.
I know, people now wants to throw me tomatoes and show me their workflow suggesting that mine is wrong.
I am sure we all agree the following as the correct LWF.
Gamma and LUT panel set to 2.2 but input and output to 1.
Affect color & material on.
In vray color mapping set gamma to 2.2 if we want to bake the correction
or
gamma set to 2.2, don’t affect color and sRGB on if we want to keep it linear.
Then load only the diffuse map with an input of 2.2
Now, I honestly think this is wrong. This is not working my renders are not photorealistic they look fake. I don’t get contact shadows, so we all try to repair with AO (that imho is the FAKE). As well the reflections goes way too strong as I switch them on. Hard to control. So we all try to save it in post. But it lacks something. A render in 2.2 takes less time because it makes less sampling. Check the evermotion gallery: beautiful images, great but imho none of them are photorealistic, maybe the.1 %
I have the feeling that the maths behind it is wrong. In my opinion the only 2 elements that needs to undergo a 2.2 treatment are the diffuse maps and the light falloff. Obviously when you switch on the sRGB button the image looks brighter so the light falloff gets modified and I think this is right. Then we need to counteract with the diffuse map to make them look properly if not they will be washed out. Fine for me. But all the rests gets corrected / brighten up as well and we don’t wan’t it to be. The fresnel effect ie it gets wrong, any falloff, procedural or greyscale map gets wrong.
I have noticed that until I am working with the only diffuse materials the renders looks fine, let’s say believable. But when I start to add reflection it’s no more right.
So I have tried to switch off the secondary bounces / I have used the reinhard for a while but still don’t get there. As well in post we try to get back some contrast bringing the gamma to 1.8 or drawing S curves but then the color shifts or get too much saturated.
I am trying to fix something that is difficult to define but I’m sure it is not correct.
I have tried softimage and now I am running some comparison tests between VRAY and MRAY (never been a fan / user) and I think there is something wrong in the way VRAY manage the LWF mathematically. I mean if you use the same workflow for both (using input and output at 2.2 this time) you’ll see the difference. With MRAY if you play with burn / midtones / shadows you get the render you look for. With VRAY you won’t. You’ll open it in photoshop / nuke and tried to get close.
As well, a question that I’ve already posted but didn’t get an answer. Is there a way to reproduce at 100% the exact color mapping in Nuke. What is the math behind the reinhard in example. Is there a way to render linear and then achieve the exact identical render you got from vray with a reinhard .5 ? I know it is probably a curve but I’d like to know the math and be able to recreate it.
Soon I’ll be back to the old good gamma 1 and exponential..
About your 1.8 workflow: do you load the texture as well in 1.8 or just set the gamma in color mapping?
Giacomo.
well the color swatches in MRay is floating number 0-1, while with VRay it uses the default max color selectors of 0-255. For VRay to match that you have to use floating numbers also. Use a VRayColorMAP
Contact shadows are nothing to do with your gamma setup afaik, you need to be looking at your GI settings to achieve correct contact shadows, and yes AO is a cheat; so use brute force with AO turned off instead.
hmm, do you still expose your camera to a .255 white vraycolor like lele did eons ago?
Thanks for the reply,
Menomonki, yes with a vraycolormap I’ll have a floating input I can have better control of dark values and reflection. but what else?
Macker, I know brute force is the solution but I don’t have a super renderfarm and have to stick with irrmap. can’t deliver noisy render or wait too long.
Kimgar, I didn’t find the link to Lele test you’re referring to. Can you give some more infos? if you mean how do I start lighting, I usually start with an override material white (230 or 240). Never pure white. is this what was it about?
thanks,
giacomo
Imho, the only thing wrong with LWF today is the same issue as I “complained” about back when LWF first appeared as something new on these forums, and that is that the usable range of colours is effectively halved, and as some have pointed out here, the severe effects on dark colour fidelity. This combined with monitors not really being calibrated for gamma 2.2 or 1, as some people like to think (I’m finding 1.5 closer to what I want on most monitors), forcing us to readjust in post. That being said, the “old” way of doing things also resulted in post adjustments, mostly the other way, upping gamma, screwing with light falloff ranges etc. is in no way better. With LWF, we at least have mathematically consistent outputs.
Where it gets confusing for me is the 240 for white or what have you. If a white wall has a 60% reflectance value (it does according to Vlado in another thread) then “white” is roughly 153. I think this what an above poster meant when they said half of the colors are essentially removed from Vray, and I don’t disagree there.
It would be nice if there was 1 definitive thread that laid it all out. But there’s like 5, and none of them agree on anything making the entire process that much more confusing.
I’ve found the thread (quite long). It seems to be dated. I wonder how many of you do actually use this workflow. I’ll give it a go. I understand that to avoid burnings and keep the render visually exponential the rule is never go over a diffuse & reflection value of 128 to leave some room for the reflection. The question is: don’t I end up having an underexposed image? What to do then? Curves in psp I guess? I’d really like to know and see some renders of people using this method that as Lele explain it’s not scientific but has got its reasons to be. Anyway that could help my lack of definition. A GI obtained from a 240 with has less falloff than a GI obtained bouncing on a 128 white (I’d say grey). I’d like to know if Peter Guthrie, Bertrand Benoit and Alex Roman they do use it…
One of the best tool I’ve recently tried is the LWF thick on the color mapping. I know the same Vlado discourage from using it but to me it gives a more predictable response to my reflection. Materials behaves exactly as in gamma 1 but the diffuse and light are corrected. As well the fog color and the fresnel has a better response in my opinion meanwhile with the “official LWF” it seems like that those value works exponentially so to fine tune a reflection you need to go crazy in between the small values of 1 - 5 while changing values from 100 to 255 it gives you a blown out reflection anyway.
Just to be redundant, I’ve seen one of the master mentioned above using the LWF thick with the old dark color swatches…
x Trixian: I tried my best to be 2.2 compliant with a calibrated Eizo monitor
Giacomo.
Brett are you really have set output to 2.2 ? With 1.8 in reinhard and output to 2.2 you make double burn.
not if he saves from the frame buffer to jpg and the like since he just burns it into the image
but yes to exr etc will have double gamma
the lele thread is here : http://www.chaosgroup.com/forums/vbulletin/showthread.php?36359-VRay-Sun-Sky-and-Physical-Camera-video-tut-Bonus-Script!/page30&highlight=lele+255+linear
but i guess you already found that
it’s pretty long yes, actually 5 years long. with 2.2 vraygamma i still find this method gives me the most contrast, but it is awkward and not very dynamic in a working environment.
i’m just curious if people are still using this method with .255 output?
hmm… I setup otput to 2.2, and rainhard to 1.8 and when saving from VFB to jpg it’s double.