Congratulations on the baby Mike!<div><br></div><div><br></div><div><br></div><div><br></div><div>[Sorry if I am breaking this mailing list's etiquette]</div><div>-Sid<br><br><div class="gmail_quote">On Tue, Jan 24, 2012 at 4:14 PM, Mike Farnsworth <span dir="ltr"><<a href="mailto:mike.farnsworth@gmail.com">mike.farnsworth@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Thanks, Brecht! I saw that this went in and the shadow bug got fixed,<br>
but I apologize for not saying anything when that was all happening.<br>
I have a good excuse, though, my wife was having a baby on Friday<br>
morning. =) (Mom and baby are home and in great shape.)<br>
<br>
So, related to this patch was the whole HDR-range texture issue...I<br>
think I'm going to take a crack at that and see if I can come up with<br>
a way to allow full-float textures for certain ones. It really only<br>
makes sense for emission (reflectances should always be zero to one),<br>
so I don't want to add overhead to all texture sampling. I'm sure<br>
there's a way to do it.<br>
<br>
-Mike<br>
<br>
On Sun, Jan 22, 2012 at 5:28 AM, Brecht Van Lommel<br>
<div class="HOEnZb"><div class="h5"><<a href="mailto:brechtvanlommel@pandora.be">brechtvanlommel@pandora.be</a>> wrote:<br>
> Problem should be fixed now, made a mistake in tweaking the patch. The<br>
> memory errors I think will also be fixed, was a separate issue with<br>
> nodes for which a fix was also committed.<br>
><br>
> Brecht.<br>
><br>
> On Sat, Jan 21, 2012 at 11:06 AM, Constantin Rahn <<a href="mailto:conz@vrchannel.de">conz@vrchannel.de</a>> wrote:<br>
>> Hi,<br>
>><br>
>> i have tested the importance sampling with one of my current scenes (Blender<br>
>> 2.61.43676, Win 7 64). But I have the problem, that there is no shadow with<br>
>> activated imp. sampling.<br>
>><br>
>> Rendering without importance sampling (250spp, CPU):<br>
>> <a href="http://www.vrchannel.de/blender/Test_is_off.jpg" target="_blank">http://www.vrchannel.de/blender/Test_is_off.jpg</a><br>
>><br>
>> Rendering with importance sampling (~50spp, CPU), only activated the "sample<br>
>> as lamp" (512) for the BG:<br>
>> <a href="http://www.vrchannel.de/blender/Test_is_on.jpg" target="_blank">http://www.vrchannel.de/blender/Test_is_on.jpg</a><br>
>><br>
>> Maybe it has to do with my node setup. I am using two different BG images,<br>
>> one JPG as visible background and one HDRI for the lighting with different<br>
>> power.<br>
>> The node setup:<br>
>> <a href="http://www.vrchannel.de/blender/Test_is_nodes.png" target="_blank">http://www.vrchannel.de/blender/Test_is_nodes.png</a><br>
>><br>
>> I get no specific "importance sampling" massage in the console. But there<br>
>> are some "Memoryblock free: attempt to free NULL pointer".<br>
>> I can't send my scene, will try to reduce the problem to a simpler scene.<br>
>><br>
>> Greetings and happy blending<br>
>> Conz<br>
>><br>
>> Am 20.01.2012 06:00, schrieb Mike Farnsworth:<br>
>><br>
>> Okay, one more update. This version hopefully fixes an issue where<br>
>> any changes to nearly any setting were causing a re-render, because I<br>
>> wasn't checking for opportunities to skip the background light sync.<br>
>> This version seems to trigger fewer progressive render updates, but<br>
>> still updates at the right times.<br>
>><br>
>> Brecht, let me know when you think this might make it into trunk.<br>
>><br>
>> Thanks,<br>
>> Farny<br>
>><br>
>> On Tue, Jan 17, 2012 at 10:59 PM, Mike Farnsworth<br>
>> <<a href="mailto:mike.farnsworth@gmail.com">mike.farnsworth@gmail.com</a>> wrote:<br>
>><br>
>> I've updated the patch to include the task split (max of 16384 shader<br>
>> evaluations per task), as well as turn off the background<br>
>> sample_as_light switch by default to not adversely impact performance<br>
>> for those who haven't set a custom environment. I don't know of any<br>
>> other outstanding issues.<br>
>><br>
>> -Farny<br>
>><br>
>> On Tue, Jan 17, 2012 at 6:18 AM, Brecht Van Lommel<br>
>> <<a href="mailto:brechtvanlommel@pandora.be">brechtvanlommel@pandora.be</a>> wrote:<br>
>><br>
>> Great work! I'll do a proper review of it later.<br>
>><br>
>> Regarding the task size, there's a function to split up tasks in the<br>
>> device code. It's used in device_opencl.cpp, task_add, if there's a<br>
>> CUDA issue with maximum size being exceeded a similar thing could be<br>
>> done there.<br>
>><br>
>> Brecht.<br>
>><br>
>> On Mon, Jan 16, 2012 at 6:16 PM, Mike Farnsworth<br>
>> <<a href="mailto:mike.farnsworth@gmail.com">mike.farnsworth@gmail.com</a>> wrote:<br>
>><br>
>> Someone tried the patch and had a crash when using an importance<br>
>> resolution greater than 1023. (It's really just eating it with a<br>
>> cuda_assert(cuLaunchGrid(cuDisplace, xblocks, 1)) when launching the<br>
>> shader task.) I had the same issue with my GTX 260, and I thought it<br>
>> was just because that was an old(er) graphics card. On my GTX 460 it<br>
>> doesn't seem to have any issues with the importance resolution. If<br>
>> the GTX 560 Ti is crashing too, though, then I should probably to do<br>
>> something about it.<br>
>><br>
>> The issue is that when running a cuda task, you can't have 65536<br>
>> blocks in any dimension. We use a block size of 16, and with a<br>
>> resolution of 1024x1024, that results in 65536 blocks in the X<br>
>> direction. The near-term workaround is to keep the resolution under<br>
>> 1024, perhaps limited in the UI.<br>
>><br>
>> The long-term solution is that I need to break up the background<br>
>> evaluation task into, say, chunks of 128x128 and go through them one<br>
>> at a time in order to cover the whole background.<br>
>><br>
>> -Farny<br>
>><br>
>> On Sun, Jan 15, 2012 at 11:35 PM, Mike Farnsworth<br>
>> <<a href="mailto:mike.farnsworth@gmail.com">mike.farnsworth@gmail.com</a>> wrote:<br>
>><br>
>> This patch includes a full implementation of importance sampling for<br>
>> environments, including the MIS solution we talked about. It adds two<br>
>> items for the user to control: on/off switch for whether to treat the<br>
>> environment as a light (off is old behavior), and a resolution control<br>
>> for the importance map. Here are a few notes about the patch:<br>
>><br>
>> * For environments that are a solid color (or very slowly changing),<br>
>> turning on the env light machinery is overkill, and will actually<br>
>> result in reduced performance. For anything complex (textures, sky,<br>
>> noise, etc) it produces noticeably less noise for the same number of<br>
>> passes, so it should be a solid advantage.<br>
>><br>
>> * Any HDR environment textures are chopped down to SDR range by the<br>
>> texture upload to device right now (nothing to do with my patch, of<br>
>> course); this reduces the effectiveness of this new environment<br>
>> sampling. If we can fix that, the env light sampling will have an<br>
>> even greater effect, not to mention potentially improved visuals in<br>
>> general. Just sayin'...<br>
>><br>
>> * The patch includes a change where the displacement device function<br>
>> now outputs to a float4 buffer, not a float3 buffer, because on my<br>
>> older GPU that was causing completely screwed up sampling of the<br>
>> background when building the importance map. So, displacement and<br>
>> shader tasks now have a slightly larger output buffer, but it now<br>
>> works on both my older GPU (GTX 260, sm_13) and my newer GPU (GTX 460,<br>
>> sm_21).<br>
>><br>
>> * There is a commented #define that allows dumping the sampled<br>
>> background and both parts of the importance map to /tmp for debugging<br>
>> purposes. Those sections of code can be removed, but they could come<br>
>> in handy in the future.<br>
>><br>
>> I'm going to post some before/after images to <a href="http://morecycles.blogspot.com" target="_blank">morecycles.blogspot.com</a><br>
>> tonight, so you can see the difference in quality. I apologize for<br>
>> this taking longer than I had anticipated. I had left in a<br>
>> __int_as_float(f) where I should have just had an (int)f instead, and<br>
>> it manifested itself with a zero PDF for the environment light, just<br>
>> causing noise. It took a while to dig through it and find it.<br>
>> Brecht, let me know if you have any issues with the code, and I'll try<br>
>> to turn around any suggestions you make ASAP so this can be included<br>
>> in 2.62. I understand the deadline is the 22nd?<br>
>><br>
>> -Farny<br>
>><br>
>> _______________________________________________<br>
>> Bf-cycles mailing list<br>
>> <a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
>> <a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
>><br>
>> _______________________________________________<br>
>> Bf-cycles mailing list<br>
>> <a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
>> <a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
>><br>
>><br>
>><br>
>> _______________________________________________<br>
>> Bf-cycles mailing list<br>
>> <a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
>> <a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
>><br>
>><br>
>><br>
>> _______________________________________________<br>
>> Bf-cycles mailing list<br>
>> <a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
>> <a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
>><br>
> _______________________________________________<br>
> Bf-cycles mailing list<br>
> <a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
> <a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
_______________________________________________<br>
Bf-cycles mailing list<br>
<a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
<a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
</div></div></blockquote></div><br></div>