PDA

View Full Version : Radiosity Cache - Animated/



devin
11-08-2013, 02:04 PM
Am I missing something or does the LW Group keep changing the way radiosity cache works, esp. when baking animation cache for a scene?

In my experience, it used to bake one scene that defaulted to "radiosity.cache'. Then it started appending a number to that one file that represented the total number of frames cached. With the official 11.6 release, it is now writing out a new cache file for each increment and appending that frame number to each file.

Am I off base? I can't find any documentation or threads covering these changes. I'd appreciate any insight.

Thanks,
Devin

RebelHill
11-08-2013, 02:19 PM
Its been changed to write out a file for each cached frame now, it also includes changes to how interpolated animated works... Frames will now be "blended" across a step distance to reduce shadow flickering. With step of 1, its really no different to how it was before, but try a step 10, and ull see the shading "crossfade" in the intermediate frames.

Its still far from a solid method, and if animated objects arent moving pretty slowly, then the crossfade effect is very obvious, but, its something I guess.

devin
11-08-2013, 11:17 PM
Good to know and thanks for the reply. Are you aware of any documentation covering this? Do we now just point the cache path to the base file name (e.g. radiosity.cache for radiosity.cache.010 - radiosity.cache.600)? My farm software, SquidNet, is now hanging up when trying to use radiosity cache but I'm running the previous build and not the latest full release version. Time to update.

Thanks again.

jwiede
11-09-2013, 04:13 AM
Good to know and thanks for the reply. Are you aware of any documentation covering this?

I really hope they cover this kind of "output file detail" in depth in the upcoming revised documentation! Such info is sorely needed for anyone using distributed rendering, and the old docs were rather poor regarding such details.

Chrusion
09-16-2014, 06:54 AM
Forgive me for reviving this thread, but the new material is directly related and why have more threads than needed for the same topic?

Several posters have written about the unsuitability of animated cached radiosity or its 'farmability.' I have had neither of these issues with any version of LW. However, after over a dozen successful anim cached rad. renders over my 4 node BNR 5.x controlled farm, my new scene using the same exact setup (duplicate scene copied to new content dir, then modified to replace main object, save filename and path, radiosity cache name, etc., but same GI environment and settings) is rendering each cached rad. frame as desired (same as F9 in Layout), but all others in between much darker, as if there's no radiosity.

The first thumb below is frames 59 (dark) and 60 (light) - click to view anim gif. The second is GI settings and cache. To repeat, every 10th frame is light (desired), all others are dark.

Any ideas why? Again, I'm just taking a successful scene, changing out the multi-layered machine LWO, the render file and path, and the cache name... everything else in the scene is the same (required files/folders from prior content copied to new content dir structure). There are NO errors in the BNR log, no "cannot find file... plugin... etc." in the LWSN console window, so I'm at a loss.

124298

124299

Chrusion
09-16-2014, 11:00 AM
Investigating further... The same light/dark effect was happening in Layout with F9 renders of any 10th frame and any other frame.

Just as a pure guess, I unchecked the unseen by rays and unseen by camera flags for my image mapped GI sphere. It's scaled so it's outside my sky and ground objects set to unseen by radiosity. Since the Clear Radiosity Cache button is grayed out for whatever reason when using Cache & Animation together (Is it because of how the new caching system is separate files now whereas it was a single file?) I manually had to delete all the previous cache files in the radiosity folder. I then rebaked the radiosity of a few frames and did F9 test renders of frames 100 and 101. The non-10th frames were the same luminance as the 10th frames (remember, using step of 10 for the cache).

SOOOO... it looks like setting unseen by rays & camera on an object that's the dominant source of GI illumination is seriously boogered in LW 11.6.3!