flythroughs?

[janams]

Hi there,

I am having the same problem, the server is re_calculating FG when using network rendering. I am still on max 8 (we do have four of max9 licenses but IT are taking time with installing it) and I've tried everything you suggested in this thread: calculating FG for the firts frame - saving it on network, unchecking RE_build option , checking use file and locating saved FGm on the network, in the explorer right clicking on the FGm file, going to Properties and in General in Attributes checking Read only option. After all this Server is still calculating FG....

I don't know what else to do.. As far as I can see the only thing you've mentioned was the "read/write option enabled" which Max8 does not seem to have. Could it be why the FG is re-calculated every time?

 

If I don't use network render and I save the FGm and tick use file, the second time I render the same scene it skips FG calculations and jumps straight to Rendering..

 

Thank you,

 

jana

[Ernest Burden III]

Now that FinalRender2/C4D has released SP2 and with it some needed stability and bug fixes, perhaps you can use that? Or do you need to stay in Max?

 

If so...

 

I would think you would want to set up a simple skylight fake with a sun and don't bother with GI for all the phases except the final one. Better yet, why not use NPR animation for all the '%' phases? It can be setup to render lightning fast and looks great and makes the point of 'work in progress', not everything is calculated and raytraced and GI'ed.

 

Also, it would be great if you could establish a set of 'keeper' camera moves so that each phase is rendered with them. That way you can do a fade animation later to show the process. Once you get to the final, you would add other paths as needed.

[janams]

OK I think I know what the problem is. In Max8 FG is view dependant, so even if I save it to a map it'll still calcuate on every frame if camera is moving. However, it adds it's calcuations to the pre-existing map, and it references that map before a new calculation. So more frames = less FG/frame time.

 

Hope this helps

[Justin Hunt]

Thats right, its only adding more samples where needed. Thats one area where Max9 has made good progress.

 

FG is view dependant, but Photons aren't (to a degree) With Max 8 you will get better and faster results by using a combination of Photons and FG. Photons for the main GI and then FG to clean up.

 

Sit on IT to get Max9 up and running you will be well pleased.

 

JHV

[janams]

Thank you Justin for coming back.

IT should be installing max 9 sometimes next week...if not I'll be back on their backs - nagging...

 

jana

[nisus]

Hi all,

 

Anyone of you tried to reload the FGM with different setting?

In max 8, I used to render a FGM with ID (ideal diffuse) on FG500, bounce 9... next re-use the file (no freeze) with FG1 bounce 0...

 

I haven't tested this in m9 yet, but it saved my ass many times ,-)

 

rgds,

 

nisus

[Justin Hunt]

Cant say I have done that, Surely if you havn't set the "reuse " then the 0 bounce calculations will over write the 9 bounce calculations.

 

Currently I am creating the FGM @ 1/4 the res of the final high res images, freezing the FGM and then rerendering @ full res. Giving me a major time saving.

 

What advantages does your methode give?

 

JHV

[Quique]

Hi there, I thought I would chime in since I'm in a situation where I'm reworking an animation I created. My understanding is that the FGM is resolution dependant. It may not matter in most scenes, but my animation is done in grayscale which definitely shows the tiniest splotches. I'm attaching 3 different images with the FGM created at 1/4 resolution, 1/2 resolution and full resolution.

 

One tip I have read earlier was to run the renderings at full resolution, but with minimal sampling (min 1/61- max 1/64) at a certain interval number of frames (every 10 frames or so - depending on speed of camera), so that the FGM could be calculated before 'freezing' and used in the final rendering.

 

My initial grayscale animation can be seen on my website: http://www.enriquecamacho.com under the animation section. That was created with GI and FG (Max 8), but this time, I'm going to try solely FG and AO (Max 9). I'm trying very hard to get my render times down, since the revised animation will be twice as long, but am still having trouble.

 

Quique

[Justin Hunt]

What are your render times, and can you post a grab of your FG settings.

 

Off the top of my head those should be done in under two minutes a frame

 

JHV

[Quique]

Hi,

 

My render times (with the frozen FGM) were the same at around 45 seconds. Rendering out the actual FGM definitely varied with the lower res FGM calculations taking about 50 seconds and the full res FGM calculation taking about 4 minutes. My FG settings are basically draft mode with 2 light bounces.

 

The 45 second rendering time is acceptable, but when I get into camera views w/lots of glass, then my times will easily quadruple, which is where it hurts. Not to mention, these settings don't have high sampling values - I think I'm using something like min 1/4 max 4.

 

Quique

[Justin Hunt]

Try Low,

 

FG point density 0,8

Rays 150

Interolate 30

 

Max Ray depth 20

increase your refractions to be relevant to the scene (how many layers of transparent objects +1)

 

Your rendertimes will increase, but the quality will be vastly better.

 

Is you Max ray depth high enough for the scene? (not the FG but for raytracing)

 

AA of 1/4 4 is fine for tests but for finals I would go up tp 4,16 which will also improve on the quality.

 

JHV

[nisus]

Cant say I have done that, Surely if you havn't set the "reuse " then the 0 bounce calculations will over write the 9 bounce calculations.

 

Currently I am creating the FGM @ 1/4 the res of the final high res images, freezing the FGM and then rerendering @ full res. Giving me a major time saving.

 

What advantages does your methode give?

 

speed. extreme increase of speed. nice smooth FG500 at the speed of FG1... useful with lots of 3d-trees... BUT as far as I know (for now), this is only max8.

 

rgds,

 

nisus

[Quique]

Justin,

 

I tried a render with the FG setttings you described. My rendering did turn out better, but at the same time, the creation of the FGM took about 35 minutes including rendering. The rendering itself (with a frozen FGM) took about the same time as the others, just under 50 seconds. (Sampling Min1/4 Max1)

 

I do wish I could increase the sampling to Min1 Max64, but just increasing it to 1-4 takes about 2 minutes for a rendering. For an animation, I can't really afford more time than that per frame. Ideally rendering times would be half of that, but then I would need to increase my hardware specs I'm sure.

 

I did review the Raytracing depth, and it is set to 5. I set it to 2 just to see what happens and the render took only about 13 seconds less and definitely had some artifacts. Perhaps I need to look into other settings to get my rendering times down. I appreciate the suggestions you have given me and hope to learn some more through these forums.

 

Thanks,

 

Quique

[Justin Hunt]

With animations it is best not to skimp on the AA. Too little and you will get horrid crawling edges and noise all over the place.

 

Try dropping the no of rays until you get an acceptable balance of time and quality.

Up the FG multiplyer to 2, to compensate for the lesser samples

 

JHV

[Quique]

Justin,

 

I agree with you about not using enough AA for animations, as I learned with this project the first time around. Concerning your suggestion, are you referring to the number of Rays per FG(or Raytracing depth)? Also, when you increase the FG multiplier by 2, does that not simply 'brighten' the FG solution? Thanks for clarifying,

 

Quique

[Justin Hunt]

Yes I am only talking about FG samples, by increasing the multiplyer, it causes the sample to travel further than the physically correct inverse square rule, thus giving the apperance that the scene is brighter.

 

I am going to contradict myself now. I was thinking last night about how you can get your rendertimes down and decided that you may be better off going with Fakeocity and AO. Seeing that this is a grey scale model you don't need to worry about colour bleed. The AO will give you the GI look, and a good light rig will negate the need to do FG.

 

Your render times will be insanely quick, even with good AA settings and there will be time saved by not haveing to run the animation twice.

 

With the Arch and Design materials AO is a simple switch of a button. If you are using Max8 then make a Oren-neylar blinn or anistropic shader and drop the AO into the diffuse level slot. Set the AO distance to something that is relevant to the scene, ie smaller than half the floor to ceiling hight and change the samples to 20 or 30. This will give the darkening with out the need to composit.

 

JHV