Created
April 20, 2011 08:01
-
-
Save anonymous/930661 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
[07:15] _samo_ hi morning | |
[07:16] condar hi _samo_ | |
[07:16] _samo_ hi there | |
[07:19] _samo_ DarkTide: ping | |
[07:19] DarkTide pong | |
[07:23] _samo_ DarkTide: hi good night | |
[07:24] _samo_ DarkTide: I wonder what's your opinion about this: http://www.yafaray.org/community/forum/viewtopic.php?f=15&t=4049 | |
[07:25] DarkTide my opinion? | |
[07:26] _samo_ yes, what do you think its the source of the slowdown | |
[07:26] _samo_ 4 times slower in path tracing is a huge difference | |
[07:29] DarkTide i know the source _samo_ | |
[07:29] DarkTide the MIS sampling strategy changed completely | |
[07:29] DarkTide since it was farely incomplete for glossy materials | |
[07:30] _samo_ Darktide, but if i render the same scene with shinydiffuse instead of glossy, the difference remains | |
[07:31] DarkTide yes, the whole sampling strategy was changed | |
[07:31] DarkTide if you look at the resulting images more than the times you'll see that images are much cleaner with fewer passes | |
[07:31] _samo_ DarkTide: I noticed that | |
[07:32] _samo_ DarkTide: the first pass produces already a cleaner result | |
[07:32] DarkTide yes that's the issue | |
[07:32] DarkTide pt is supersampling from the begining | |
[07:32] _samo_ ok then I guess with yafaray 012 fewer passes are needed then? | |
[07:33] DarkTide i yes | |
[07:33] DarkTide also the results are consistent with pm and dl | |
[07:33] DarkTide which isn't the case on 0.1.1 | |
[07:34] _samo_ ok that's an answer thanks | |
[07:34] DarkTide yw | |
[07:35] _samo_ I guess we should explain that when we release the rc and the stable, since everyone is going to complain | |
[07:35] DarkTide i can try to remove the supersampling on pt and see if the times and the results get a better balance | |
[07:35] DarkTide yes | |
[07:35] _samo_ you don't need to, only that we should explain users what to expect now | |
[07:36] DarkTide ok | |
[07:36] DarkTide but know that there is still some room for optimization, this changes were made with consistensy and correctness in mind not optimization | |
[07:37] DarkTide not much room but some | |
[07:37] _samo_ DarkTide: if you think the new sampling is better, I'm ok with it | |
[07:38] _samo_ in fact the result is good with yafaray 012, so it is only a question of using fewer passes | |
[07:40] DarkTide a simple test to really see the difference is to render a glossy sphere with only bg texture | |
[07:42] _samo_ it will produce a less noisy result in the first pass? | |
[07:42] DarkTide 0.1.1 has no bg reflection on glossy | |
[07:42] DarkTide using pt i mean | |
[07:45] _samo_ ok I'm going to test that | |
[07:50] -->| nickon (~nn@d54C0E153.access.telenet.be) has joined #yafaray | |
[07:50] _samo_ DarkTide: I should enable 'use background' in yafaray 011, right? | |
[07:51] DarkTide nope that will lit the sphere much like ibl does | |
[07:51] DarkTide so is not exactly the same effect | |
[07:52] _samo_ ok | |
[07:54] _samo_ DarkTide: what kind of background should I use to test that? | |
[07:54] DarkTide any texture bg | |
[07:54] DarkTide in fact iirc any bg | |
[08:01] _samo_ DarkTide: yes I see differences | |
[08:01] condar _samo_: http://dl.dropbox.com/u/4113119/reflection011.png | |
[08:01] condar http://dl.dropbox.com/u/4113119/reflection012.png | |
[08:02] _samo_ DarkTide: even if the background does not emit light, the glossy material in 012 is reflecting the blue sky | |
[08:03] _samo_ condar: yes I get similar results | |
[08:05] _samo_ http://img821.imageshack.us/g/yafaray011.png/ | |
[08:07] _samo_ the sampling is completely different in yafaray 012 | |
[08:08] DarkTide now render that same scene in 0.1.1 with dl and then do the same with 0.1.2 | |
[08:12] _samo_ ok let me try that | |
[08:14] condar http://dl.dropbox.com/u/4113119/dl011.png | |
[08:14] condar consistency have a price | |
[08:22] _samo_ http://img600.imageshack.us/g/yafaray011dl.png/ | |
[08:27] DarkTide now you've seen what i ment about consistensy and the sampling strategy change | |
[08:29] _samo_ this is what I get with the background not emitting light : http://img851.imageshack.us/i/yafaray012dlnbgl.jpg/ | |
[08:29] _samo_ in yafaray 011 I get a black image | |
[08:30] DarkTide this is suomi's scene with 8 samples on pt | |
[08:30] DarkTide http://dl.dropbox.com/u/6595937/8-samples-pt.png | |
[08:30] _samo_ sorry I'm wrong, | |
[08:30] _samo_ in direct lighting, yafaray 011 delivers a correct result | |
[08:31] condar so in theory 0.1.2 needs or less samples or less pases to get same result? | |
[08:31] _samo_ I mean with no background lighting I can see the background still reflected on the sphere | |
[08:31] _samo_ but in path tracing yafaray does not reflect the background, only the lighting, I'm right? | |
[08:33] DarkTide _samo_: yes you are right | |
[08:33] DarkTide condar: that is correct | |
[08:33] _samo_ DarkTide: one thing I was not sure about when comparing suomi's result is the fact that I'm not sure whether the yafaray 012 result is using the same bump strenght | |
[08:35] _samo_ that's why I could not compare the sampling either | |
[08:35] DarkTide the bump algorithms was changed | |
[08:35] DarkTide -s | |
[08:36] _samo_ ok, but they using the same strenght when the use the same values? i mean yafaray 011 compared to yafaray 012 | |
[08:36] _samo_ *are they | |
[08:37] _samo_ *they | |
[08:38] _samo_ DarkTide: also can you produce the same scene in yafaray 011 please? | |
[08:38] condar 1min 56 instead of 5 min 30 sec | |
[08:38] condar http://dl.dropbox.com/u/4113119/16samplesPT.png | |
[08:38] _samo_ with 8 samples | |
[08:38] condar but only 16 samples | |
[08:39] DarkTide condar can you do the same for 0.1.1 please i don't have 0.1.1 at hand | |
[08:39] condar and the quality is pretty much the same | |
[08:39] subcomandante oh noooo, the bump algorithm noooooooo | |
[08:39] DarkTide subcomandante: don't worry _bert fixed it | |
[08:39] subcomandante j/k | |
[08:39] DarkTide | |
[08:39] condar _samo_: same scene but 64 samples 5min 30 sec | |
[08:39] condar http://dl.dropbox.com/u/4113119/012.png | |
[08:39] DarkTide then... | |
[08:39] DarkTide i hate you!! | |
[08:39] DarkTide hehehe | |
[08:40] _samo_ condar can you try the same sampling in yafaray 011, I want to compare results instead of render times | |
[08:40] condar 1 minute | |
[08:41] -->| sandkign (4e85a702@gateway/web/freenode/ip.78.133.167.2) has joined #yafaray | |
[08:41] sandkign hi | |
[08:42] DarkTide sandkign? | |
[08:42] sandkign i was just wondering if there is difference if i use one large piece of glass intersecting with walls and simulating windows or should i make each window separately | |
[08:42] sandkign ? | |
[08:42] DarkTide you misspelled your nick | |
[08:42] sandkign DarkTide: aye? | |
[08:42] sandkign nevermind :d | |
[08:42] DarkTide or aren't you sandking? | |
[08:42] sandkign i use this stupid webchat as after IT reorganization they won't let me have irc client | |
[08:42] DarkTide hehehe | |
[08:43] sandkign so back to the glass - is using one piece of glass in the wall ok ? | |
[08:43] DarkTide well i'm guessing the large glass doesn't make any difference as lokg as it is embedded in the wall | |
[08:43] DarkTide long* | |
[08:43] sandkign or should glass panes not interfere physically for correct calculations? | |
[08:43] sandkign ok | |
[08:43] condar sandkign: but it's better to keep windows separately for alrge projects... believe me | |
[08:44] condar *large | |
[08:44] sandkign i remember it was quite important when modelling water in glass to have proper caustic effects | |
[08:44] sandkign condar: yeah, but it's a side prop now | |
[08:44] DarkTide for realism is bettes since you can hcange the angles of the windows a bit and get a realistic appearance | |
[08:44] sandkign i'm detailing windows of the main thing | |
[08:45] sandkign oh, i didn't thought about changing arc trick - thanks | |
[08:45] DarkTide damn i need to sleep some time soon | |
[08:45] sandkign i use normal map for glass imperfection sometimes | |
[08:45] -->| Hyper-Core (~lol@h174.117.31.71.dynamic.ip.windstream.net) has joined #yafaray | |
[08:45] sandkign DarkTide: don't waste time on sleep | |
[08:46] _samo_ hi subcomandante | |
[08:46] DarkTide i'm trying but morpheus call is too strong | |
[08:46] subcomandante hola _samo_ | |
[08:46] DarkTide btw subcomandante: chiao fratello!!! | |
[08:46] DarkTide damn | |
[08:46] subcomandante | |
[08:46] condar hi subcomandante | |
[08:46] DarkTide ciao* | |
[08:46] DarkTide hehe | |
[08:46] sandkign anyone by chance knows this tutorial that floated about a month ago when a chap is doing evening viz of office building by the street? he detailed some light tricks there and i can't find it . it wasn't yafaray/blender related as i remember | |
[08:48] condar _samo_: http://dl.dropbox.com/u/4113119/011pt64samples.png | |
[08:48] DarkTide condar: and 16? | |
[08:48] condar rendering | |
[08:48] DarkTide thx | |
[08:49] condar http://dl.dropbox.com/u/4113119/011pt16samples.png | |
[08:49] _samo_ condar, really? why the reflections are so different | |
[08:49] condar because is glossy.. | |
[08:49] DarkTide bg reflection i'm guessing | |
[08:49] DarkTide is a white bg? | |
[08:49] condar *it's | |
[08:50] <--| Hyper-Core has left #yafaray | |
[08:50] condar yep | |
[08:50] DarkTide then there is the difference | |
[08:50] condar the 16samples version for 0.1.2 took the same time than a 64samples 0.1.1 version | |
[08:50] condar and the quality is close | |
[08:51] DarkTide yep | |
[08:51] DarkTide can you render 0.1.2 16 samples with a black bg after the 16 samples 0.1.1 version? | |
[08:51] condar so test 0.1.1 vs 0.1.2 with the same ettings is pointless | |
[08:51] DarkTide please? | |
[08:51] condar 1 minute | |
[08:52] DarkTide thank you very very much, you are the best tester we have on the room right now | |
[08:52] DarkTide ... | |
[08:52] DarkTide the only one i ment | |
[08:52] DarkTide hahahaha | |
[08:53] condar hehehehe | |
[08:54] condar so ... 0.1.2 is not slower than 0.1.1.it make better sampling | |
[08:54] condar *makes | |
[08:55] condar man... 4 am and I cannot control the keyboard | |
[08:55] condar http://dl.dropbox.com/u/4113119/012blackBG.png | |
[08:57] DarkTide so glossy is still missing pieces in 0.1.1? | |
[08:57] DarkTide can you run a dl test? | |
[08:57] condar 0.1.1? | |
[08:57] DarkTide i'm sorry for being such a pita but i don't have 0.1.1 | |
[08:57] DarkTide yes please | |
[08:58] condar back to 0.1.1................ | |
[09:01] condar http://dl.dropbox.com/u/4113119/011DL.png | |
[09:02] condar yep... 0.1.2 FTW | |
[09:02] DarkTide ?? | |
[09:04] condar http://dl.dropbox.com/u/4113119/011DLblackBG.png | |
[09:06] condar ok... I need to sleep... cu later | |
[09:06] |<-- condar has left freenode (Quit: Leaving.) | |
[09:08] _samo_ this is my results with bump maps disabled : http://img541.imageshack.us/g/yafaray011pt16.png/ | |
[09:09] DarkTide that's quite a difference | |
[09:09] _samo_ less noise in yafaray 012, it is also reflecting the background unlike yafaray 011 | |
[09:17] _samo_ DarkTide: in MIS sampling strategy, was does MIS stand for | |
[09:18] DarkTide Multiple Importance Sampling | |
[09:20] _samo_ darktideok thanks for you time to explain all of this | |
[09:20] _samo_ Darktide: ok thanks for you time to explain all of this | |
[09:21] DarkTide | |
[09:21] DarkTide yw _samo_ now i go to bed | |
[09:21] DarkTide c u tomorrow | |
[09:24] _samo_ cu DarkTide |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment