This is it! Thanks a lot :) I guess I will just fix it by either setting the instanced VAOs to non-conflicting index ranges or reset the divisors before draw calls (for now).
On Sunday, October 4, 2015 at 6:21:32 PM UTC+2, Floh wrote: > > This could be the same bug that popped up on the WebGL dev group here: > https://groups.google.com/forum/#!topic/webgl-dev-list/0Omhxnp3_CE, which > points to a bug in ANGLE when instanced_arrays is used together with vertex > array objects. > > It doesn't look like a trivial bug though (back when I still used VAOs I > didn't have trouble with the instanced_array extension in simple demos). > > Cheers, > -Floh. > > Am Sonntag, 4. Oktober 2015 15:29:50 UTC+2 schrieb Wladimir Kröker: >> >> Hey, >> I've been playing around with emscripten and tried instancing but ran >> into a problem in windows browsers: >> I'm rendering 1 blue square with glDrawArrays() and 6 yellow squares with >> glDrawArraysInstanced(). Unfortunately the result >> isn't quite what I expect it to be. At least in Windows browsers: >> >> Expected: >> >> <https://lh3.googleusercontent.com/-QS9y7wRn-1g/VhEaqyVtijI/AAAAAAAAAGw/hGVh2mGYX3w/s1600/Expected.png> >> >> Result in Windows browsers (Firefox, Chrome, Opera): >> >> >> <https://lh3.googleusercontent.com/-V7kosdauX1I/VhEayeK2V2I/AAAAAAAAAG4/AKt20p8pi84/s1600/Result.png> >> >> >> I get the expected result if I start the program as an "exe" on Windows >> or on Linux (executable on the OS and in browsers). >> I tested it with the 1.34.1 emsdk and the portable version on linux. >> >> Note: It doesn't matter on which OS I create the .html. It just won't >> show me the expected result in the mentioned windows browsers. >> >> The test code can be found here: >> https://github.com/compix/Emscripten-Instancing-Test >> I tried to keep it as minimal and readable as possible with a few error >> checks. >> >> The cause of the problem seems to be the glVertexAttribDivisor(0, 1) in >> InstancingTest.cpp line 156. I didn't have any problems without it. >> >> I am using 2 attributes for the instancing case: "in_sPos" as the >> instanced attribute with divisor 1 and "in_pos" as a non-instanced >> attribute. >> If I change the order of the attributes in the test code I get the >> expected result. But this doesn't really fix the problem because >> 1. The order shouldn't matter. >> 2. It will cause problems if more geometry and different vertex layouts >> come in. >> >> Maybe there is a problem in my code but I checked it multiple times and I >> don't think it's a coincidence that it just doesn't display the expected >> result in windows browsers. >> >> I'd much appreciate if someone could take a look at it. >> >> Regards >> Wladimir >> >> >> -- You received this message because you are subscribed to the Google Groups "emscripten-discuss" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
