light map photon map, load save both please help

Post your tutorials and help questions in this forum
Locked
katsuy
Posts: 7
Joined: Sat Jan 06, 2007 3:57 am
Location: canada

light map photon map, load save both please help

Post by katsuy »

hi I am trying to sort what path to use for what purpose
could some one shed some light on the correct work flow for say a exterior animation with only camera moving...

my guess is it would be lightmap, shared and load... but if the camera flys to the other side of the building(hidden face to frame 0 of animation) it should be both, wrong?
If this is the case what is the correct way to save the initial gi solution. is photon estimate enough? to you have to use photon mapping....


also i read somewhere that if you choose photon map(as opposed to light map) in this situation you shoot all photons required hence requiring less fg rays at render time. Does this means that it will render the entire sequence faster at the cost of the first frame being slower??? Or is my understanding off...

finally what is cache irradiance exactly do?

forgive my multi question post
k
Attachments
varibles.jpg
varibles.jpg (65.76 KiB) Viewed 3552 times
jure
Posts: 2142
Joined: Thu Jun 02, 2005 6:53 pm

Re: light map photon map, load save both please help

Post by jure »

Let me first explain Shared file GI modes when rendering sequence of frames:

SAVE: this mode will always save FG data to GI file but will not reuse it for next frame. It will just keep adding new data with each frame. This setting is only good for testing in precomputed mode where you want data to get saved before you move to FG testing where you want to use that saved data.

LOAD: This mode will load FG data from file and compute any missing FG data that it cannot find. So if some pixels don't have enough data in GI file it will compute new FG data but it will not save this new data to the file. This mode is good for testing FG settings. You use save when you do precopmuted render and then change to load mode when you do FG tweaking.

Both: This mode will check for available data in GI file and if found it will use it for computing FG. If data is not found it will compute new FG data and then save it to shared file so it can be reused for consequitive frames.

There will be another mode added in next OB release which will be similar to LOAD but it will not compute any new FG data at all. This will be good to split render to computers on render farm.

Now to your questions:
katsuy wrote:hi I am trying to sort what path to use for what purpose
could some one shed some light on the correct work flow for say a exterior animation with only camera moving...
my guess is it would be lightmap, shared and load... but if the camera flys to the other side of the building(hidden face to frame 0 of animation) it should be both, wrong?
If this is the case what is the correct way to save the initial gi solution. is photon estimate enough? to you have to use photon mapping....
I recomend using BOTH mode for camera flythroughs. It doesn't matter if it's Lightmap or photonmap - both should work fine. The only difference is that in Lightmap mode KRay will do additional preprocessing by going through whole animation quickly and shoot photons.
katsuy wrote: also i read somewhere that if you choose photon map(as opposed to light map) in this situation you shoot all photons required hence requiring less fg rays at render time. Does this means that it will render the entire sequence faster at the cost of the first frame being slower??? Or is my understanding off...
No this is not true. The only difference between lightmap and photonmap is that in Lightmap mode photons are shot from the direction of the camera and in photon map from the light sources. Lightmap is better because it won't waste photons in placed you don't see. For example if you have many rooms in a house animation lightmap mode will only shoot photons along camera path. Photonmap mode will shoot photons from all lightsources in house so every room will get photons even though it may not be visible in animation at all. This means alot of wasted photons.
katsuy wrote: finally what is cache irradiance exactly do?
Cache irradiance is a technique of computing data only on important places. This means that when "cache irradaince" is ON kray will search for places where change of light is most prominent (Corners, close to objects, shadows etc.). It will put alot more samples in those places as opposed to large flat areas.
This will be described in more depth in upcoming manual.
- Jure
katsuy
Posts: 7
Joined: Sat Jan 06, 2007 3:57 am
Location: canada

Post by katsuy »

thanks jure for the prompt reply

starting to make more sense now.

for some reason i was under the impression that save, load and both were independent from FG....

Photonmap mode will shoot photons from all lightsources in house so every room will get photons even though it may not be visible in animation at all. This means alot of wasted photons.
so when exactly would we use this? when we bake to texture i guess...

now was there a tutorial for that somewhere?

cache irradiance...ahhh...should be called something like refine irradiance, no?

k
User avatar
phile_forum
Posts: 177
Joined: Wed Aug 16, 2006 11:12 pm
Location: UK
Contact:

Post by phile_forum »

OK, so say I have a scene and I want to take four shots of it from four different positions using four cameras.

Can I use 'Both' mode to help speed up the rendering of the 2nd, 3rd & 4th shots?
jure
Posts: 2142
Joined: Thu Jun 02, 2005 6:53 pm

Post by jure »

katsuy: I use Ligtmaps only. Photon maps can still be used sometimes though. They seem to work well with small lightsources like lamps and such. Or for example a scene with a laser beam behind where the source of the beam is not directly visible. Because the source of light is extremely small Lightmap photons that come from camera will be unable to catch it. But Photon maps would work very well since it would fire photons from the source itself.

Cache irradiance is a technical term describing this kind of calculation. But yes for easier understanding for average user it could be changed.

Phile: Yes you definitely can use Both. Just animate camera to those four positions and use "render sequence" button.
- Jure
katsuy
Posts: 7
Joined: Sat Jan 06, 2007 3:57 am
Location: canada

Post by katsuy »

Both: This mode will check for available data in GI file and if found it will use it for computing FG. If data is not found it will compute new FG data and then save it to shared file so it can be reused for consequitive frames.
ok so if say, i must interrupt a 90 frame sequence, is the gi used at frame 45 good to render back frame 10 or because of its progressive nature this gi is no longer "good" to render the previous frames.

in some of the animations posted here there was a suddent burst, was this caused by a "obsolete solution"?


Cache irradiance is a technical term describing this kind of calculation. But yes for easier understanding for average user it could be changed.
:D :oops:
jure
Posts: 2142
Joined: Thu Jun 02, 2005 6:53 pm

Post by jure »

ok so if say, i must interrupt a 90 frame sequence, is the gi used at frame 45 good to render back frame 10 or because of its progressive nature this gi is no longer "good" to render the previous frames.
Since GI data was allready computed for frames up to 45 it can be used with no problem for rerendering the frames again. But you must be aware that if you changed light or surfaces the GI data that will be loaded from file will not be correct so artifacts may appear.

in some of the animations posted here there was a suddent burst, was this caused by a "obsolete solution"?
This happens for example when you render animation with shared GI file till a certain frame and then restart render from there with a new GI file. Old and new GI render therefore will not match since GI solution cannot be exactly the same even though you haven't changed anything.
- Jure
Locked