Hey Vlado, great tutorial, that’s really helped me out alot
Just two questions for you… my apologies if I’m going over old territory!
1. What setting did you use for the IRmap sequence? Medium-animation? or something quicker?
2. Am I right in noticing that once the light cache has been calculated, the IRmap phase get calculated one hell of a lot faster?
3. (the bonus question) I just wanted to clarify… It is possible to use incremental add to current map in conjunction with backburner without any problems (ie flicker)? Does backburner handle frame allocation ok without max saving any files? Reading your post above, I get the feeling you are suggesting that you need to send different sections of the animation to each render-node.
This is my first post here, so I wanted to also say a big thanks for writing such a great package!
2. Am I right in noticing that once the light cache has been calculated, the IRmap phase get calculated one hell of a lot faster?Faster compared to what? It is certainly a lot faster than using QMC GI for the secondary bounces.
3. (the bonus question) I just wanted to clarify… It is possible to use incremental add to current map in conjunction with backburner without any problems (ie flicker)? Does backburner handle frame allocation ok without max saving any files? Reading your post above, I get the feeling you are suggesting that you need to send different sections of the animation to each render-node.Yes, you can render different sections on different nodes, but you must merge the resulting irradiance maps manually with the irradiance map viewer in the end, before rendering the final animation.
This is my first post here, so I wanted to also say a big thanks for writing such a great package! You are welcome
I am still confused about that last part of the tutorial. Why could you do without secondary bounces when you have glossy reflections in the scene? Or am I misunderstanding things here :roll:
you can have the lightcache saved with the irradiance map. If done so, you can actually just turn off secondary bounces which will save memory as you wont have to load the lightcache.
If you have the lightcache help precompute the glossies, then youll need to go ahead and load the lightcache when rendering.
To have all the multiframe inc. maps coming from the same machine, I tried DR, thinking it would be faster.., the first IR went well, but the second one could not log the other machines in the job… (doing it this way would avoit the manual assembly of the final map, well, I think :? )
This can be done, for example, by creating a geosphere directly in the camera viewport or a Tape object and using it to visualize the sample size. For our scene, a sample size of about 2.0 seems to be good enough (the scene is in Generic units).
Didn’t understand that part. Took out a tape measure and still having problems trying to identify what was measured.
It can help you determine how large is a single sample from the light cache on the average - that is, the size of the patches that you see in the unfiltered light cache. You don’t want these to be too big, as you will loose detail in the lighting, neither too small as this will make it too noisy. In Screen mode for the light cache, the size is adjusted automatically, but since we are using World mode in the tutorial, you have to determine the right size for the samples based on the specific scene.
i think to have a problem in the part II for my personal proyect, when I calculate the irradiance map (medium-animation) in primary engine I have to stay set the second engine to lightcache (from file: .vrlmap 28 MB) but it is very very slow calculate it I think (resolution 180x135->10 minutes 1 frame and i need a resolution of 768X576->25 mimutes*300 frame ). is it normal?. If I set the second engine to none or QMC the Imap calculate is speed (11 sg) but then it is not correct, not?
i have a AMD64 3200+, 1GB Ram, WinXP SP2, 3dmax 6, vray 1.47.03. My scene is a small exterior with a plane modifier Greeble (plugin)
Why are you using, as AA, adaptive subdivision instead of qmc?
I’ asking cause everyone says that adaptive qmc is better. However I’ve been using it and got very noisy results with glossys, and turning global subdivisions up makes rendering too slow. So I thought may be to start using adaptive subdivision again.
Just a little clarification (I have always been used to Imap and QMC for animations - this Lightcache option has opened up a whole new can of worms…)
1) In Part II, do we leave the secondary engine set to Lightcache, and have that Lightcache loaded from the file we saved in Part I? In this way, the Irradiance map is being calculated from the Lightcache.
Do I understand this correctly?
2) We are producing a massive animation of a business park. There are about 30 large office buildings. The camera flies around the whole park and also goes into two or three of the buildings (it is proving to be quite a challenge!). The whole model is built in mm, so you can imagine that the scene size is many tens of thousands of units across. Should I leave the Lightcache set to Screen mode rather than World?
3) I may be pushing things just too far with this same animation. As I understand it, the Subdivs setting for the Lightcache in Flythrough mode is a number of samples that is spread *evenly* throughout the whole of the animation path. In the above scenario with 30-odd buildings on a massive business park, my Subdivs setting would need to be many thousands (?). Am I right? In your tutorial, you set it to 3300 - this is for a very small scene in comparison to the business park I am working on. This suggests a figure of 30000+ Subdivs! I assume this would either take for ever and a day to calculate or crash my system.
Is this where the ‘brute force’ method comes into play, and if so, the dreaded ‘flicker’?
Do you think an animation like this is even feasible with Vray?
In part two the lightcache is indeed used by the ir map calculation to speed things up. This is the same in a single frame, the LC is always calculated first so that IR map can use this info.
In your scene I think it is best to break the anim up into multiple scenes so you don’t have to calculate the LC for the whole anim at once. You would indeed need insane amount of subdivs especially when you enter the buildings too.
You can still use world scale (which should be better otherwise samples in the distance will be much larger than those close to the camera). But don’t set it to 0.01 if you work in mm… You don’t want samples to be 1/100th of a mm small Also here, breaking up in scenes would be better so you can use smaller samples in your interior vs exterior.
I used it for an interior scene, but it was a pretty easy scene. I don’t know if it will make much of a difference in exteriors (because there isn’t much secondary bounces anyways). But for interiors it is definatly way faster.
Another thing: the tutorial mentions:
“Make sure that the timeline animation range matches the range which you want to render”
How do I do this? Basically, I have quite a few cameras of varying lengths from 0 to 90 seconds. If I am looking at one specific camera with a range from 20 to 30 seconds, how do I achieve this? Is it just by setting Time Output/Range in the Render/Common box?
The reason why I am thinking this doesn’t work is because when it builds the LC, I can see elements in the samples from much earlier in the cameras path than I wish to compute. Does this make sense?
The LC animation mode should only be used for 1 camera I think.
If cam one moves from frame 300 to 1200, set the visible part of the timeline from 300-1200 and render the LC for that cam. Do the same for the other cams and save them all to different files so you can load the appropriate one when you will do the IR map pass for each cam.
I guess you know how to make a certain range visible? Just press the time configuration button next to the frame number field and set start and end time.
). is it normal?. If I set the second engine to none or QMC the Imap calculate is speed (11 sg) but then it is not correct, not?