Fstorm render

[Moriah]

What i noticed from archviz guys trying out FStorm (and some personal experience), is that:

1. No need to setup, since it's "unbiased", you dont have to worry about settings/technical side of things (maybe GI clamp if you need).
2. Glare/Bloom is very simple to setup (realtime too) and gives that touch of "realism" that you sometimes lack with Vray because of it not being so simple to setup and not realtime.
3. Lighting behavior seems different for some reason, both with single light sources and single HDRI, maybe it's the GI, but artistically it looks "better"/"different".
4. Very fast for a GPU renderer but lacks a lot of features. Vray on the other hand is not as fast but is way more complete in GPU.

I even asked to one of the gurus there what was his opinion on Vray and he said "there's too much settings and setup needed, i'm not a technical guy, i'm an artist, with Fstorm i just have to click render and it's already photoreal even with the lack of features it has". - I don't agree with this, nowadays Vray is pretty simple, but you see what i'm trying to say...

So from the technical point of view actually there might not be that many differences, i don't know and i can't comment much on that, but from an artist's perspective, it's easier to achieve a photoreal look indeed with less fiddling around.

Of course this is for "simple" archviz scenes, because other than that you can't do much with it right now, specially if you're into VFX work or something that needs some degree of compositing. But then again, there's no real renderer that is a master of everything, but when the job needs to be done, people always go back to Vray.

[vlado]

I even asked to one of the gurus there what was his opinion on Vray and he said "there's too much settings and setup needed, i'm not a technical guy, i'm an artist, with Fstorm i just have to click render and it's already photoreal even with the lack of features it has". - I don't agree with this, nowadays Vray is pretty simple, but you see what i'm trying to say...

That once an idea gets into an artist's head, it's very difficult to take it out? Yes, V-Ray used to be more complicated to set up, but that's not the case anymore. It will take a bit of time for that thought to sink in.

I can't really agree that lighting behavior is different - all render engines compute pretty much the same thing. There are details in mostly in clamping, limited light bounces and other "optimizations" (f.e. Corona renders are by default more contrast-y because it clamps GI heavily, RedShift is even more aggressive sometimes to the extent that the final render has little to do with the real world), but overall they produce similar results.

Best regards,
Vlado

[savage309]

I agree on the real time lens effects. We are working on having that, but it will take a while
Unbiased usually can be translated to "unsmart". We have the BF/BF option. I don't see how having no cached GI implementation can be a feature.
V-Ray, especially RT GPU has no complex options or whatsoever and it is very easy to use. From the feedback we have got so far from various artists, RT GPU is not slower by any means compared to whatever GPU raytracer.

Best,
Blago.

[Moriah]

Glad to know an improved lens effects system is coming! Also i might do a simple scene comparison with vray and fstorm and try to see where these differences come from hopefully.

What about what he claims to be a modified GGX BRDF? "FStorm material uses unique developed glossy brdf model. This model has less noise level and looks better than modern GGX brdf model." Do you think this might influence the overall look of the scenes?

[savage309]

It might be the case that there is a better shader, but the approved way of arguing about such claims is by publishing a paper (like ours for SIGGRAPH this year and the ones in the docs.chaosgroup.com), which can be checked by multiple independent authors. It happens occasionally somebody to have a different shader/light/whatever, because the original one was too hard to implement.

Best,
Blago.

[jstrob]

I've heard the same argument about any number of render engines - Maxwell, Octane, Corona, Arnold, iray etc. However under similar lighting conditions and similar materials all these engines produce results that are more or less identical to V-Ray (and I know this because I've checked, many times). Don't get me wrong, if you find that FStorm works better for you, by all means use it, I just don't accept the "more photoreal" argument as valid.

Best regards,
Vlado

I was not not impressed enough by Maxwell (except maybe those single flash photos once), Corona, Octane or Arnold (maybe except the movie "Gravity") to think I really need to learn how to use them, but I think there is a small plus with fstrom. But I will test it just for fun cause it is really far from being usable in my work. For that V-Ray is way more advanced and versatile. My render farm is all CPU based and it's still the best way to have predictable results for animation. I need render element, cpu based rendering and all the shaders vray has as well as all the pipeline tool that comes with vray (open vdb support, vray instancing etc). I also need the absolutely perfect support we get here from Vlado and chaosgroup! (BTW, Good luck to Arnold's users now that Autodesk owns it!). A renderer needs more than a photoreal single frame button. I haven't seen any animation from Fstorm yet.

And yes for sure I will try to see if I can reproduce some of the Fstrom Render in V-Ray and some compositing. For stuff like glare etc, it's always better in comp anyway.

[Nicinus]

Some quick comparisons between Fstorm and Redshift.

I don't get it, in these Redshift looks much better at a third of the time?

[Donfarese]

I don't get it, in these Redshift looks much better at a third of the time?

That's what I thought Also, Redshift is much better if you ask me.

[Nicinus]

That's what I thought Also, Redshift is much better if you ask me.

Could you please add a quick test with Vray RT as well?

[joconnell]

I was not not impressed enough by Arnold (maybe except the movie "Gravity")

Just as a thing, I just finished working on the new star trek film and it was interesting to see how a film that big is put together. On an asset similar to the bits you're seeing in gravity, you'd have a team of maybe 15 or twenty people working on the space station model for about a year. Likewise the characters would have had a massive team working on them for quite a while using very, very heavily layered shaders which are almost impractical for a small studio to render. The typical render node in one of those studios might have 64 or 128 gigs of ram and they're happy to let a single 2k frame run for 12+ hours. A betrand benoit scene or even something like grant warwicks current still life has been worked on for weeks and weeks - it doesn't matter that much about the renderer you use, it's the talent of the artist and the amount of time they've got to refine things!

[Nicinus]

Just as a thing, I just finished working on the new star trek film and it was interesting to see how a film that big is put together. On an asset similar to the bits you're seeing in gravity, you'd have a team of maybe 15 or twenty people working on the space station model for about a year. Likewise the characters would have had a massive team working on them for quite a while using very, very heavily layered shaders which are almost impractical for a small studio to render. The typical render node in one of those studios might have 64 or 128 gigs of ram and they're happy to let a single 2k frame run for 12+ hours. A betrand benoit scene or even something like grant warwicks current still life has been worked on for weeks and weeks - it doesn't matter that much about the renderer you use, it's the talent of the artist and the amount of time they've got to refine things!

That's just the thing, most of us (I would assume) don't have that kind of resources and need as effective tools as possible.

[jstrob]

Just as a thing, I just finished working on the new star trek film and it was interesting to see how a film that big is put together. On an asset similar to the bits you're seeing in gravity, you'd have a team of maybe 15 or twenty people working on the space station model for about a year. Likewise the characters would have had a massive team working on them for quite a while using very, very heavily layered shaders which are almost impractical for a small studio to render. The typical render node in one of those studios might have 64 or 128 gigs of ram and they're happy to let a single 2k frame run for 12+ hours. A betrand benoit scene or even something like grant warwicks current still life has been worked on for weeks and weeks - it doesn't matter that much about the renderer you use, it's the talent of the artist and the amount of time they've got to refine things!

Yes you're absolutely right that a movie like Gravity is not good example to compare renderer. They would have use V-Ray and probably could have get the same result with much less load on the render farm (saving a few millions in hardware).

and very interesting insight about your work on Star Trek. i worked a bit on some movies too and I didn't really to spend a year on a small detail that no one will ever notice and that's why I prefer working in advertising for TV, youtube or or short movies.

[glorybound]

Just a thought. Maybe, big ticket productions get a lot of things free. I could imagine that software and hardware are giving in exchange for a mention. You know, the rich pay far less for things than the poor. This could be why they used FStorm.

[jstrob]

Just a thought. Maybe, big ticket productions get a lot of things free. I could imagine that software and hardware are giving in exchange for a mention. You know, the rich pay far less for things than the poor. This could be why they used FStorm.

you meant "Arnold"

[glorybound]

Yes, whichever. I couldn't image why a big budget would not go with V-Ray. All these others seem good, but none seem to have anything over V-Ray, other than price.