Fstorm render

[LudvikKoutny]

It's the GI clampling at 1.0 and the 8 diffuse bounces by default which make it contrasty.
the LC will distribute light across 100 bounces, making for a substantially brighter render, under identical color mapping conditions (those include Gamma, btw.)

I do not think it's a good idea to generally assume that parameters across different renderers translate 1:1. I assumed GI clamp in Fstorm is re-scaled, so that it doesn't reflect actual intensity that GI is being clamped to, in image units, but rather re-scaled to something artistic friendly. It makes often sense, to have 1.0 as a value that represents 100% of the recommended value.

Another reason for me to think that is that I've seen how renders with secondary rays clamped to 1.0 really look, and Fstorm output is nowhere near that. You would see a lot of secondary reflections missing. And of course, even if there was extreme amount of clamping and insufficient of bounces, it would still not make nearly as much difference as on Bobby's renders. What happened there is the usual syndrome of automatic translation, where many things translate differently or get lost in translation completely.

It doesn't really make much sense to attribute all the difference to a few ray depth and clamping render setting, expecting that if they are raised higher, that Fstorm render will magically end up looking same as V-Ray. And speaking of defaults, if we compare apples to apples, V-Ray still to this day defaults to 3 indirect bounces for Brute Force secondary GI, and all the V-Ray materials still default to (these days IMHO insufficient) 5 reflection and 5 refraction bounces, so Fstorm can't really be blamed much here

But of course, as usually, I decided to back up my arguments with some proof, so I established current conditions:

Fstorm:
- Increased Max Depth to 12
- Disabled all tonemapping settings to make image 100% linear
- Disabled any glare and bloom

V-Ray:
- Enabled global ray depth override to 12
- Set secondary GI to Brute Force with 12 bounces

So here is V-Ray with Max Ray Intensity of 20 compared to Fstorm with GI Clamp at 1:

Looking at secondary reflections... Fstorm looks a bit brighter, so apparently GI clamp doesn't directly translate to Max Ray intensity.

Now, here is V-Ray with Max Ray Intensity of 20 next to the Fstorm with GI Clamp at 20:

Again, a big difference in favor of Fstorm in nearly identical scene.

So I decided to multiply V-Ray's MRI by 20, resulting to value of 400, and compare it to Fstorm's GI Clamp of 20

Looks quite close

And in the end, I decided to make sure by raising Fstorm's GI clamp to 400:

And got an actual caustics, even though noisy, which V-Ray at MRI 400 completely clamped.

So to wrap it up, I am not implying one renderer better than another, but I don't think one can just jump in saying: "Doesn't matter it makes pretty pictures, it's bad because it has that one number in render settings at that value, and it's not the same number we have, we have bigger number, so we win". So while obviously Fstorm doesn't clamp that much (although it does in general, but so does V-Ray), even if it did clamp the hell out of our images, at the end of the day, how pleasing it looks to the eye is what matters the most

[LudvikKoutny]

And as usually, I am attaching scene files for anyone to verify/reproduce the results.

[jstrob]

For sure it would be great to have a better lens effect in v-ray. I never used it cause it looks worse than what I can do in post. But if we had a glare like fstorm that look even better that what I can do in post, I would surely use it.

[glorybound]

Setting to CUDA just sits the VFB on Waiting for image data... and never moves. This has never worked for me.The VFB gets stuck and I have to kill MAX to do anything else.

Oh god yeah that's not going to work, I hope you at least set it to Cuda or you was actually using CPU RT. Fstorm is actually already setup really good out of the box to just hit render. RT GPU you need to set a few settings.

[^Lele^]

I do not think it's a good idea to generally assume that parameters across different renderers translate 1:1. I assumed GI clamp in Fstorm is re-scaled, so that it doesn't reflect actual intensity that GI is being clamped to, in image units, but rather re-scaled to something artistic friendly. It makes often sense, to have 1.0 as a value that represents 100% of the recommended value.

It's my job to measure this stuff, and one i have already completed exhaustively for the current release.
Feel free to explore it yourself, however.

[mitviz]

same here, RT gpu i guess needs more powerful cards, takes ages on mine to start while cpu starts right away

[Donfarese]

Setting to CUDA just sits the VFB on Waiting for image data... and never moves. This has never worked for me.The VFB gets stuck and I have to kill MAX to do anything else.

That really should not be happening, especially since you have a GTX 980 TI with more than enough vram for that scene. Do you have the latest drivers? and are you using some sort of software like EVGA Precision X for controllingcard, that sometimes can mess things up.

[Donfarese]

Same with you Mitviz, you have a decent card, unless your loading large scenes. Both you and Glorybound should test a simple scene first (Teapot). If that works fine then there might be to big of scene or something in the scene not compatible with Vray.

[glorybound]

FStorm worked without a hitch. I have used VRay RT CPU for setting up my HDRI's, but that's about the extent of it. My driver is current and yes, I have the NVIDIA Experience installed.

[M.Max]

Setting to CUDA just sits the VFB on Waiting for image data... and never moves. This has never worked for me.The VFB gets stuck and I have to kill MAX to do anything else.

I don't think that should happen too with your 980 TI ! Just to make sure you have used VrayRT in production mode and not the Active mode ?! I don't know if that should happen, but I found the active shade to be slow and hard to deal with (GTX 970 here ) but in production mode it works better.

[glorybound]

I set it as my production render (CUDA) and after 30 seconds, I got this. It was way over exposed, so I used VFB's exposure controls. When it hit render, after LC, it seems to just pause. I see the VFB update 2-3 times, which is what you are seeing below, but it seems there are no progress.

[glorybound]

1 minute with everything default. This is not RT, it's V-Ray Adv, with all defaults. Not to shabby.

[mitviz]

1 minute with everything default. This is not RT, it's V-Ray Adv, with all defaults. Not to shabby.

[ATTACH=CONFIG]31576[/ATTACH]

your testing RT or just the scene now?

I never actually ran RT in production mode, what are the benefits over RT in active shade really?

[glorybound]

RT in production cleaned up fast through LC, however, it basically slowed to a halt when it actually started to render. I then went back to V-Ray normally and hit render with the defaults, and it cleaned up just as fast in LC and then kept rendering and got pretty clean pretty fast. Long story short, for me, regular V-Ray is faster.

[PIXELBOX_SRO]

Getting little OT guys...