FryRender - Atrium

[Devin Johnston]

Yes Nicole announced last week that "a change in the licensing policy is near. We will announce it in a couple of weeks".

[u.biq]

What are Fry’s policies’ on licenses, how many do you get with one copy and do they count cores instead of processors?

Hi Devin,

 

Quote:

10-computer floating license with no CPU limit as a single network hardware dongle

 

Cheers

[Devin Johnston]

Does 10 computers mean 10 CPU's or 10 Cores, and what does no CPU limit as a single network hardware dongle mean?

[Ernest Burden III]

Does 10 computers mean 10 CPU's or 10 Cores, and what does no CPU limit as a single network hardware dongle mean?

 

I should know this since I recently bought Fry. I think it means 10 computers--10 IPs. However many CPUs, however many cores in those CPUs. The 'protection' is a toothlike glowing red dongle. So I suppose you can plug it into any machine to render from. I'm sorry to continue the 'supposing', but I think that's what U.big meant.

 

EDIT:

I looked this up on the Fry forum. As written recently by Chema, the owner of FeverSoft:

 

"A license of fryrender unlocks 10 computers, and there is no core or CPU limit. It unlocks IPs in fact. 10 IPs."

[CHE]

Ernest,

 

What is your opinion about Fryrender so far? Have you used it in production yet? How different is the work flow from Maxwell's?

[Ernest Burden III]

What is your opinion about Fryrender so far? Have you used it in production yet? How different is the work flow from Maxwell's?

 

I am the wrong person to ask. I put Maxwell down early on and never learned much beyond the fact it was a soap-opera.

 

With Fry, I've pushed a few buttons so far. I've only had it since Monday, and had too much work to be able to try it out much. So far so good. It all makes sense, though the docs are still very thin. Chema answers questions promptly and honesty on the Fryum to make up for the lack of documentation so far. From what Fran and others have done it seems like glass and other important things actually work nicely. Materials are easy to apply and set up, it will read native C4D materials to some degree. And all else being equal, its put out by a decent, trustworthy company, and that gives it the edge.

[u.biq]

I should know this since I recently bought Fry. I think it means 10 computers--10 IPs. However many CPUs, however many cores in those CPUs. The 'protection' is a toothlike glowing red dongle. So I suppose you can plug it into any machine to render from. I'm sorry to continue the 'supposing', but I think that's what U.big meant.

 

EDIT:

I looked this up on the Fry forum. As written recently by Chema, the owner of FeverSoft:

 

"A license of fryrender unlocks 10 computers, and there is no core or CPU limit. It unlocks IPs in fact. 10 IPs."

 

All correct

nothing to add

 

I just think NL will at least match this, but who knows, they can be so fucked up

[u.biq]

Have you used it in production yet? How different is the work flow from Maxwell's?

I take the liberty to answer to that, i speak for myself, of course

 

Yes, i've used it in production since i've bought it. You can watch these archviz productions on my website or in the fry forum

 

Workflow is quite similar to maxwell, it is just working better, and you have no studio cause it's perfectly integrated in your app

 

my .02

[Devin Johnston]

The big factor between Maxwell and Fry seems to be speed, it's been said that Fry is faster but I don't know if a direct comparison has ever been done. Has anyone ever tested this?

[AJLynn]

Ernest or Fran or somebody-

 

Can you do a test render and find out whether Fry has the MLT-caustics-problem? That is to say, when you see a surface in a mirror or through glass that has caustics hitting it, you don't see the caustics. Or would that be an NDA violation?

[u.biq]

The big factor between Maxwell and Fry seems to be speed, it's been said that Fry is faster but I don't know if a direct comparison has ever been done. Has anyone ever tested this?

 

it's a bit difficult as Fry is not even in V1.0.

so a true side by side comparaison doesn't really make sense yet...even if i have my opinion, of course

[Devin Johnston]

I constantly hear that Fry is better than Maxwell so a side by side comparison seems to be the perfect way to settle the debate. I’m more interested in seeing if it’s really faster.

[CHE]

Yes, I agree. It would be great to have a little test scene we could all share and compare between Maxwell and Fryrender users in the forum.

[Ernest Burden III]

About the NDA. We are not allowed to post screenshots at all (GUIs), and images can be posted only with prior permission on FS. But they have been very good about approving images to be used off their forum.

 

If I re-install MWR I could test, and frankly I am rather curious about that caustics thing myself. I will see if I can set that up and get an answer up when I have one, images if possible.

 

Right now I'm just trying out some of the features on simple object scenes until I'm comfortable. But so far, its been pretty easy.

[Ernest Burden III]

By the way, I've already done a very important side-by-side comparison. I've got Chema vs Victor, I've got Fran vs Tom The Child God, NL vs FS. I'm happier with Fry.

[Ernest Burden III]

Can you do a test render and find out whether Fry has the MLT-caustics-problem? That is to say, when you see a surface in a mirror or through glass that has caustics hitting it, you don't see the caustics. Or would that be an NDA violation?

 

I set up a simple scene with a supplied glass material on a cone, made a spotlight to shine through it to cause a caustic on the floor. I put a double-poly sheet of glass covering most of this and have an object nearby with a mirror material. So there will be a caustic on the floor and you will see/not see it through the glass and in the reflection and both.

 

Reading the NDA I need to pass this by FRYquarters before I can post results.

 

However, I can say that I ran the test for 5 minutes and am still quite happy with my investment in Fry.

[AJLynn]

I just tried another test of this in Maxwell since I hadn't done it in 1.1. A bit hard to show the setup, but...

 

The floor, left wall and medium sized box are light gray with roughness 100. The right wall and the largest of the three boxes are mirror surfaces. The small (and very thin) box is an emitter. (Color in the render is because it's a warm light preset and I didn't take the time to balance it.) The medium box blocks the left wall from the emitter, so the only light it gets is caustics bounced off the large box. The floor gets direct light.

 

The render is SL16. 76 minutes on my Core2 Duo, which is about 4x as fast as the P4 I originally ran the test on. The first thing to resolve nicely was the light on the floor, then the caustics on the wall, then the reflection of the floor. The reflection of the caustics is showing up but took quite a while to show anything more than a few bright pixels. It probably would actually look okay afte a couple more hours but since SL16 is higher than I ever get with a print-res production render it wouldn't have been a realistic test. I'm being generous - I should have stopped it at 14.

[Devin Johnston]

Ernest have you done the same test with Maxwell yet?

[Ernest Burden III]

Ernest have you done the same test with Maxwell yet?

 

The NDA requires that we do not post comparison tests to other engines during the beta period. I haven't tested in Maxwell, but you can.

 

Here's a 3ds file with the objects and a camera:

http://www.oreally.com/temp2/caustics-test-01.3ds

 

I'm waiting to hear from Chema on posting the Fry proof-of-caustics greatness test, but in the meantime, try this in Maxwell. If it takes more than five minutes on a single core machine to know where its going (at 480x360) there is a problem.

[AJLynn]

I think that scene is a bit confusing, since th emitter can "see" floor directly - the floor gets some direct light that's not caustics. But you've probably got a better render than I do that makes more sense. Meanwhile I'm trying it with a glass teapot that's completely blocking the front of the can.

 

Here's mine, in Max9 and 3DS, in case anybody wants to see for themselves. For anybody who's wondering why this is important - this is the issue that caused the problem in Maxwell where you have a building with windows, there's sunlight, you're looking at the building from the outside, and through the window you see a very dark room.

 

This was before they added the AGS (architectural glass) material, which I think was added by members of the beta "A" team who took suggestions that some of us were making and did some outside-the-box thinking. But the problem remains for any times when you need that glass to refract, or when it's a reflection involved instead of a transparency.

[Ernest Burden III]

I think that scene is a bit confusing, since th emitter can "see" floor directly - the floor gets some direct light that's not caustics.

 

I could test a scene that you set up, just post it (now that we're ruining Ubig's thread).

 

there are some limitations in Monte Carlo integration when it comes to indirect caustics. Fry collides with those limitations and from what I've read Maxwell does too. Those are not exactly 'engine-related limitations' but 'technology-related limitations': cases that certain methods of integration can't address correctly. Raph posted some nice tests with pools to overcome those limitations in the very common case of a pool. A sunlit pool is a good 'bad case', where the caustics are produced by a deterministic refraction and also seen through a deterministic refraction.

 

What I did yesterday after reading your question:

five minutes at this size on single-core AMD 2700:

 

 

I let it run overnight:

 

 

So if this test does not show what you are curious about either describe how it should be different or post one to work with.

[AJLynn]

I just did one in Maxwell the same as that one and got the same result. Oh well, I guess that's enough of this, sorry about the thread hijacking Ubig

[numerobis]

hi Ernest,

 

nice test!

is this "ghostglas" in the foreground or "real" glas with refraction and dispersion?

[Ernest Burden III]

I just did one in Maxwell the same as that one and got the same result. Oh well, I guess that's enough of this, sorry about the thread hijacking Ubig

 

The same result as what I got in Fry?

 

Yes, its the provided regular glass I used. I haven't tried 'ghost glass' yet.

 

And Andrew, I see what you are looking for in the test model. It's not so much about the glass (it has none) but other bounced light. I agree, that could take a while to resolve.