Questions on workflow

[DanSHP]

Hi all, I've been meaning to create this thread for a while and this may be a long post, be grateful for some of your input.
I've worked in the Arch Viz sector for around 8 years now (still images interior/exterior), and wanted to just run through some questions, get your thoughts? I'm a little bit confused as things stand, I’ve read through many discussions on this forum and whilst some of it has been useful reading, it's not especially clear what path I should be following follow. Such a thing can happen when there is so much information to take in.

I would like to provide example images of my scenes but, this is difficult due to NDA' etc, so I will attempt to layout some of the head banging questions I seem to be asking myself over and over again on every job


1. Render Settings (attached)


The settings aren’t too far the defaults. I’ve always worked with Linear Multiply, 32 bit .exr’s. I adjust the noise threshold to reduce noise, sometimes to 0.005 for production level, with max sub divs at 100, I never change min.

Is there anything in these settings that stand out as incorrect or needs adjusting in regards to the above? I appreciate, it’s sometimes scene dependent, but in general???


2. Correct linear workflow



I’ve always been under the assumption that using Linear Multiply colour mapping, means an image that I save out, say an .exr or .tiff , is therefore, linear. Is this 100% true, or partially/completely false?

For example I’ve read somewhere on the forum, that to be in true linear, an image should be saved out in gamma 1.0 and then corrected later (where or how I don’t know). Along the lines of 0-1.0 float value??? This scuppered my understanding of the linear world and I feel weak. Again, I’m not sure what to make of this.

3. Workflow/post production with linear imagery (as I understand linear to be!).



I post produce in Photoshop, I hate it . I hate it even more now that you are no longer able to Filter>Camera Raw on a 32 bit .exr since I updated our PS. I am desperately wanting to move out of it, into a proper 32 bit supported environment, but I’ve struggled to find anything that offers the functionality that camera raw offers.

As I understand, linear imagery will have within it, blown out highlights. The great thing about Camera Raw in 32 bit, was the ability to take a linear image and tone down these burnt out highlights specifically, not to mention having great control over many different aspects of an image with full 32 bit depth control. This was my method.

I haven’t come across many people ever using camera raw when working with 32bit linear imagery. Question is, how else can you deal with it? When working in linear, and dealing with blown out highlights, this was the only way I was able to have some control over them. How else can you control them, outside of vray? Maybe my understanding of this is incorrect?

I honestly would just love to cut out post production in another program altogether, and be given tools in the VFB, which is I know is on the way.

4. Lighting/Camera settings



I have a collection of high quality HDRI’s from, VizPeople. I also have some of Peter Guthrie’s. I use the camera settings (attached) chart (obtained from this forum search function) to then go ahead and test lighting my scene.

Sunny day f/16, too dark on first test? No problem right? Just crank up the power of the HDRI multiplier until you are happy with the look in the VFB. This works, but again its hard to know what to tinker, for example when my scene is overexposed (burnt out). My camera settings are set to real world examples but are they wrong for this scene? is it my materials that are too bright? Is it because I’m working in linear? How do you discern what to change?


I will add some more later! If any of you could be so kind to help me out with some of this, I can actually feel confident in what I am doing again.

Cheers

[Muhammed_Hamed]

Is there anything in these settings that stand out as incorrect or needs adjusting in regards to the above?

With bucket mode, you don't need 100 max subdivs.. 32 is enough for all of your scenes and even default value of 24 is not bad.
Inspecting sample rate render element between using 32 and 100 max subdivs, you will find few tiny changed
With progressive sampler, leave max subdivs at 100 or 50.. There shouldn't be a problem with this, as Vray adds more more samples progressively whenever it needs to. But I would recommend bucket mode whenever possible as it is faster than progressive and it will give you cleaner result using the same noise threshold
Everything else should be fine. You can use min shading rate of 6 with all your scenes and use noise threshold to control how clean your render is. Usually .008 or .007 is enough for me
I tend to use value of 16 for min shading rate in GI intensive scenes, this help with cleaning noise overall and you can get away with using higher noise threshold like .009 or .01
Value of 3 for min shading rate would be nice if you have strong DOF effect in your render. You will notice the difference specially with progressive sampler
But like I said if you want to stick with min shading rate of 6 with all your scenes, it should be fine.
For GI, I use BF + LC for all of my renders.. I never use IR now as Vray next is really fast with BF and the amount of details in ares of contrast is worth a few more minutes of render time.
LC subdivs of 1500 to 2000 should be enough for most scene.

Is this 100% true, or partially/completely false?

No idea really.. I use Reinhard for all my renders and barely do I play with the burn value. Vray for Maya/Vray for Modo user here

How else can you control them, outside of vray? Maybe my understanding of this is incorrect?

People use Nuke or Affinity photo now. Personally I still use the old camera raw filter.
Sad that Adobe dropped support for 32 Bit Camera RAW :/

for example when my scene is overexposed (burnt out). My camera settings are set to real world examples but are they wrong for this scene? is it my materials that are too bright? Is it because I’m working in linear? How do you discern what to change?

Personally, I don't mind if my camera settings are not set to real world values. Sometimes I would change the multiplier of the HDRI or play with iso/shutter speed to get a nice balance. For materials, I usually don't use 100% white for reflection/diffuse colors .. Same for saturation of my colors, I don't go above a certain value..

Muhammed

[dgruwier]

Linear multiply with .exr is fine, that will for sure save as linear. Also Reinhard with default settings is just linear multiply.

The render engine, textures, compositing and file formats should all generally be kept or converted into linear as you work with it. But computer monitors need images to be gamma corrected before displaying them. The VFB does that by default (that's what the sRGB button does), and it can also be done with luts and openColorIO files, but always think of it as a preview. The image underneath is still linear. Same goes for compositing, you're working in linear, but previewing gamma corrected in the viewport.
Photoshop and After Effects and raw editors are a sort of strange middle thing, in that they generally work in 2.2 color space already, and you have to be careful to be able to preserve the linear workflow. I don't know much about how to do it in practice.

It's correct that linear images will often have values above 1, which becomes white when displayed on a monitor. You reduce the contrast and roll off the highlights with grading and color correction, but again I don't do that kind of stuff a lot, so I can't help with specifics.

Something that confused me for a long time was that if linear is the most "correct" why does it look wrong when displayed on a computer monitor? As far as I can tell, it's because computer monitors just assume the images they receive are already gamma 2.2 sRGB, and actively correct the images to be linear as they're displayed. So if you send it an image that is already "correct", it'll be corrected twice and show too dark. So a linear image, corrected to sRGB. Here's a good source to better understand all of this. It's VFX centric, so some of it probably isn't relevant to you, but the basics still apply.