PhotonGI cache + Oidn params

Use this forum for general user support and related questions.
User avatar
rickyx
Posts: 23
Joined: Sun Apr 29, 2018 5:23 pm

PhotonGI cache + Oidn params

Post by rickyx » Sat Aug 31, 2019 12:11 am

Hi!

I'm trying to reproduce the wonderful images (the door ajar) at the bottom of the page:
viewtopic.php?f=5&t=840&sid=e29e845956b ... &start=810
but I find it difficult to calibrate the parameters well.

Here is my current image and params:
Test02-Params.jpg
door ajar
Can you help me?
Thank you,
Riccardo

User avatar
epilectrolytics
Donor
Donor
Posts: 561
Joined: Thu Oct 04, 2018 6:06 am

Re: PhotonGI cache + Oidn params

Post by epilectrolytics » Sun Sep 01, 2019 6:00 pm

Hi Riccardo!

In the PGI cache settings the brute force radius scale is way too low (4mm x 8 = 3.2 cm, better 1cm x 100 = 1m).
When you make the lookup radius smaller you always need to increase the radius scale in order to compensate.
Also 4mm lookup radius is very small, I would try if 10mm works (you can also make the door thicker on the outside where it won't be seen).

If this doesn't help please share your scene file (only room with door and lights without furniture).
MBPro 15" 16GB i7-4850HQ GT750M, MacOS 10.13.6 & Win10Pro PC 16GB Ryzen 2700X, 2 x RTX 2070

provisory
Posts: 224
Joined: Wed Aug 01, 2018 4:26 pm

Re: PhotonGI cache + Oidn params

Post by provisory » Sun Sep 01, 2019 6:14 pm

Yes, try 1cm for Lookup Radius and 90 for Brute Force Radius Scale, and walls should have thickness too (al least 2x the Lookup Radius, just like the door).

User avatar
rickyx
Posts: 23
Joined: Sun Apr 29, 2018 5:23 pm

Re: PhotonGI cache + Oidn params

Post by rickyx » Sun Sep 01, 2019 11:58 pm

Attached you can find the purged file.

I have tried with Lookup 1cm and Brute force 90 and it is better but i always get "painting" effect on ceiling, walls and fine details.
Walls and door have got realistic thickness, should I increase them?

Thank you,
Ricky
master-Cache01-P.blend
(2.87 MiB) Downloaded 20 times

provisory
Posts: 224
Joined: Wed Aug 01, 2018 4:26 pm

Re: PhotonGI cache + Oidn params

Post by provisory » Mon Sep 02, 2019 7:49 am

Lookup Radius is set to 1 meter instead of 1 cm in your file.

And for parquet and wall materials you use mix of diffuse and glossy. In LuxCore it isn't needed and I think PGI doesn't work well with Mix material.
Just use Disney or Glossy material. (Disney is like Principled in Cycles, just with less features [yet, hopefully]).

provisory
Posts: 224
Joined: Wed Aug 01, 2018 4:26 pm

Re: PhotonGI cache + Oidn params

Post by provisory » Mon Sep 02, 2019 8:39 am

Some more thoughts:
- I used PGI Debug: Show Indirect for er, uh... debugging :)
- There is still a bright spot on the left side of the foreground countertop cabinet, but this is light reflected from that cabinet, so it is correct.
Attachments
PGI-Kitchen- Rickyx.jpg

User avatar
rickyx
Posts: 23
Joined: Sun Apr 29, 2018 5:23 pm

Re: PhotonGI cache + Oidn params

Post by rickyx » Mon Sep 02, 2019 9:18 am

provisory wrote:
Mon Sep 02, 2019 7:49 am
Lookup Radius is set to 1 meter instead of 1 cm in your file.
Ahh, I made a mistake in writing!

That is my result: fixed materials, from mixed to glossy.
1cm Lookup
Test03.jpg

User avatar
epilectrolytics
Donor
Donor
Posts: 561
Joined: Thu Oct 04, 2018 6:06 am

Re: PhotonGI cache + Oidn params

Post by epilectrolytics » Mon Sep 02, 2019 12:08 pm

rickyx wrote:
Mon Sep 02, 2019 9:18 am
That is my result: fixed materials, from mixed to glossy.
1cm Lookup
Test03.jpg
Looking good!

I found that in the file most materials were not included in the cache, even when setting PGI glossiness threshold to 0.05.
Only when I changed wall material to matte and floor to glossy they worked :?

My version:
Screenshot.jpg
MBPro 15" 16GB i7-4850HQ GT750M, MacOS 10.13.6 & Win10Pro PC 16GB Ryzen 2700X, 2 x RTX 2070

User avatar
rickyx
Posts: 23
Joined: Sun Apr 29, 2018 5:23 pm

Re: PhotonGI cache + Oidn params

Post by rickyx » Mon Sep 09, 2019 11:26 pm

I'm trying the PhothonGI on the same scene with a different lightning.
provisory wrote:
Mon Sep 02, 2019 8:39 am
- I used PGI Debug: Show Indirect for er, uh... debugging :)
I did but I really don't know how those spot means. Are they correct?
Debug.jpg
DebugFull.jpg
Discovering this feature is a trial/error testing...

After having a good cache I should be able to speed up animation, is this correct?

Thank you,
Riccardo

User avatar
rickyx
Posts: 23
Joined: Sun Apr 29, 2018 5:23 pm

Re: PhotonGI cache + Oidn params

Post by rickyx » Mon Sep 09, 2019 11:55 pm

A note: when you enable OpenCl on a card that does not support it, Blender crashes.
This is the log on my Dell with 2 graphic cards, Intel and Nvidia: I have the opencl version of LuxCore and Linux as operating system.

I should not press that button ;) but I think that the error should be handled,
thank you,
Riccardo

Code: Select all

[LuxRays][10.729] OpenCL Platform 0: Intel
[LuxRays][10.729] OpenCL Platform 1: NVIDIA Corporation
[LuxRays][10.737] Device 0 name: NativeThread
[LuxRays][10.737] Device 0 type: NATIVE_THREAD
[LuxRays][10.738] Device 0 compute units: 1
[LuxRays][10.738] Device 0 preferred float vector width: 4
[LuxRays][10.738] Device 0 max allocable memory: 0MBytes
[LuxRays][10.738] Device 0 max allocable memory block size: 0MBytes
[LuxRays][10.738] Device 1 name: Intel(R) UHD Graphics Coffee Lake Halo GT2
[LuxRays][10.738] Device 1 type: OPENCL_GPU
[LuxRays][10.738] Device 1 compute units: 24
[LuxRays][10.738] Device 1 preferred float vector width: 4
[LuxRays][10.738] Device 1 max allocable memory: 4096MBytes
[LuxRays][10.738] Device 1 max allocable memory block size: 3072MBytes
[LuxRays][10.738] Device 2 name: GeForce GTX 1060 with Max-Q Design
[LuxRays][10.738] Device 2 type: OPENCL_GPU
[LuxRays][10.738] Device 2 compute units: 10
[LuxRays][10.738] Device 2 preferred float vector width: 1
[LuxRays][10.738] Device 2 max allocable memory: 6078MBytes
[LuxRays][10.738] Device 2 max allocable memory block size: 1519MBytes
[LuxRays][10.739] Creating 14 intersection device(s)
[LuxRays][10.739] Allocating intersection device 0: Intel(R) UHD Graphics Coffee Lake Halo GT2 (Type = OPENCL_GPU)
[LuxRays][10.739] Allocating intersection device 1: GeForce GTX 1060 with Max-Q Design (Type = OPENCL_GPU)
[LuxRays][10.739] Allocating intersection device 2: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 3: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 4: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 5: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 6: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 7: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 8: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 9: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 10: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 11: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 12: NativeThread (Type = NATIVE_THREAD)
[LuxRays][10.739] Allocating intersection device 13: NativeThread (Type = NATIVE_THREAD)
[LuxCore][10.739] OpenCL devices used:
[LuxCore][10.739] [Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect]
[LuxCore][10.739]   Device OpenCL version: OpenCL 2.0 beignet 1.3
[LuxCore][10.739] [GeForce GTX 1060 with Max-Q Design Intersect]
[LuxCore][10.739]   Device OpenCL version: OpenCL 1.2 CUDA
[LuxCore][10.739] Native devices used: 12
[LuxCore][10.739] Configuring 2 OpenCL render threads
[LuxCore][10.739] Configuring 12 native render threads
[LuxRays][10.739] Preprocessing DataSet
[LuxRays][10.739] Total vertex count: 1684
[LuxRays][10.739] Total triangle count: 3100
[LuxRays][10.739] Preprocessing DataSet done
[LuxRays][10.739] Adding DataSet accelerator: BVH
[LuxRays][10.739] Total vertex count: 1684
[LuxRays][10.739] Total triangle count: 3100
[LuxRays][10.740] BVH Dataset preprocessing time: 0ms
[LuxRays][10.740] BVH builder: EMBREE_BINNED_SAH
[LuxRays][10.742] BVH build hierarchy time: 2ms
[LuxRays][10.742] BVH total build time: 3ms
[LuxRays][10.742] Total BVH memory usage: 147Kbytes
[LuxRays][10.742] Adding DataSet accelerator: EMBREE
[LuxRays][10.742] Total vertex count: 1684
[LuxRays][10.742] Total triangle count: 3100
[LuxRays][10.747] EmbreeAccel build time: 5ms
[LuxRays][10.749] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] BVH mesh vertices buffer size: 19Kbytes
[LuxRays][10.750] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] BVH nodes buffer size: 147Kbytes
[LuxRays][10.863] [Device GeForce GTX 1060 with Max-Q Design Intersect] BVH mesh vertices buffer size: 19Kbytes
[LuxRays][10.864] [Device GeForce GTX 1060 with Max-Q Design Intersect] BVH nodes buffer size: 147Kbytes
[LuxCore][10.875] [PathOCLRenderEngine] OpenCL task count: 524288
[LuxCore][10.876] [PathOCLBaseRenderEngine] OpenCL max. page memory size: 1556048Kbytes
[LuxCore][10.876] Compile Geometry
[LuxCore][10.876] Scene geometry compilation time: 0ms
[LuxCore][10.876] Compile 32 Textures
[LuxCore][10.876] Textures compilation time: 0ms
[LuxCore][10.876] Compile 4 Materials
[LuxCore][10.876] Material compilation time: 0ms
[LuxCore][10.876] Compile Lights
[LuxCore][10.877] Lights compilation time: 0ms
[LuxCore][10.877] Compile ImageMaps
[LuxCore][10.877] Image maps page(s) count: 1
[LuxCore][10.877]  RGB channel page 0 size: 1728Kbytes
[LuxCore][10.877] Image maps compilation time: 0ms
[LuxCore][10.877] Always enabled OpenCL code: 
[LuxCore][10.877] Compile Geometry
[LuxCore][10.877] Scene geometry compilation time: 0ms
[LuxCore][10.877] Compile 32 Textures
[LuxCore][10.877] Textures compilation time: 0ms
[LuxCore][10.877] Compile 4 Materials
[LuxCore][10.877] Material compilation time: 0ms
[LuxCore][10.877] Compile Lights
[LuxCore][10.877] Lights compilation time: 0ms
[LuxCore][10.877] Compile ImageMaps
[LuxCore][10.877] Image maps page(s) count: 1
[LuxCore][10.877]  RGB channel page 0 size: 1728Kbytes
[LuxCore][10.877] Image maps compilation time: 0ms
[LuxCore][10.877] Starting 2 OpenCL render threads
[LuxRays][10.889] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] RADIANCE_PER_PIXEL_NORMALIZEDs[0] buffer size: 18225Kbytes
[LuxRays][10.889] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] ALBEDO buffer size: 18225Kbytes
[LuxRays][10.889] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] AVG_SHADING_NORMAL buffer size: 18225Kbytes
[LuxRays][10.889] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] NOISE buffer size: 4556Kbytes
[LuxRays][10.902] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] RADIANCE_PER_PIXEL_NORMALIZEDs[0] buffer size: 18225Kbytes
[LuxRays][10.902] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Camera buffer size: 5448bytes
[LuxRays][10.902] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Normals buffer size: 19Kbytes
[LuxRays][10.902] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] UVs buffer size: 13Kbytes
[LuxRays][10.902] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Vertices buffer size: 19Kbytes
[LuxRays][10.902] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Triangles buffer size: 36Kbytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Mesh description buffer size: 624bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] ImageMap descriptions buffer size: 28bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] ImageMaps buffer size: 1728Kbytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Textures buffer size: 8704bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Materials buffer size: 592bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Scene objects buffer size: 48bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Lights buffer size: 1660bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Env. light indices buffer size: 4bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Light offsets (Part I) buffer size: 16bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Light offsets (Part II) buffer size: 16bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] LightsDistribution buffer size: 48bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] InfiniteLightSourcesDistribution buffer size: 48bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Ray buffer size: 24576Kbytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] RayHit buffer size: 10240Kbytes
[LuxCore][10.903] [PathOCLBaseRenderThread::0] Size of a GPUTask: 452bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] GPUTask buffer size: 231424Kbytes
[LuxCore][10.903] [PathOCLBaseRenderThread::0] Size of a GPUTask DirectLight: 96bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] GPUTaskDirectLight buffer size: 49152Kbytes
[LuxCore][10.903] [PathOCLBaseRenderThread::0] Size of a GPUTask State: 296bytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] GPUTaskState buffer size: 151552Kbytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] GPUTask Stats buffer size: 2048Kbytes
[LuxRays][10.903] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] SamplerSharedData buffer size: 4556Kbytes
[LuxCore][10.905] [PathOCLBaseRenderThread::0] Size of a Sample: 112bytes
[LuxRays][10.905] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Sample buffer size: 57344Kbytes
[LuxCore][10.905] [PathOCLBaseRenderThread::0] Sample dimensions: 69
[LuxCore][10.905] [PathOCLBaseRenderThread::0] Size of a SampleData: 8bytes
[LuxRays][10.905] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] SampleData buffer size: 4096Kbytes
[LuxRays][10.905] [Device Intel(R) UHD Graphics Coffee Lake Halo GT2 Intersect] Pixel Filter Distribution buffer size: 33Kbytes
[LuxCore][10.907] [PathOCLBaseRenderThread::0] Defined symbols:  -D LUXRAYS_OPENCL_KERNEL -D SLG_OPENCL_KERNEL -D RENDER_ENGINE_PATHOCL -D PARAM_RAY_EPSILON_MIN=1.000000e-05f -D PARAM_RAY_EPSILON_MAX=1.000000e-01f -D PARAM_LIGHT_WORLD_RADIUS_SCALE=1.050000e+00f -D PARAM_ACCEL_BVH -D PARAM_FILM_RADIANCE_GROUP_0 -D PARAM_FILM_RADIANCE_GROUP_COUNT=1 -D PARAM_FILM_CHANNELS_HAS_ALBEDO -D PARAM_FILM_CHANNELS_HAS_AVG_SHADING_NORMAL -D PARAM_FILM_CHANNELS_HAS_NOISE -D PARAM_ENABLE_TEX_CONST_FLOAT -D PARAM_ENABLE_TEX_CONST_FLOAT3 -D PARAM_ENABLE_TEX_IMAGEMAP -D PARAM_ENABLE_TEX_SCALE -D PARAM_ENABLE_CHECKERBOARD2D -D PARAM_ENABLE_TEX_NORMALMAP -D PARAM_ENABLE_TEX_FRESNELCOLOR -D PARAM_ENABLE_TEX_FRESNELCONST -D PARAM_ENABLE_MAT_MATTE -D PARAM_ENABLE_MAT_GLOSSY2 -D PARAM_ENABLE_MAT_DISNEY -D PARAM_HAS_PASSTHROUGH -D PARAM_CAMERA_TYPE=0 -D PARAM_HAS_CONSTANTINFINITELIGHT -D PARAM_HAS_TRIANGLELIGHT -D PARAM_HAS_ENVLIGHTS -D PARAM_HAS_IMAGEMAPS -D PARAM_IMAGEMAPS_PAGE_0 -D PARAM_IMAGEMAPS_COUNT=1 -D PARAM_HAS_IMAGEMAPS_BYTE_FORMAT -D PARAM_HAS_IMAGEMAPS_1xCHANNELS -D PARAM_HAS_IMAGEMAPS_WRAP_REPEAT -D PARAM_HAS_BUMPMAPS -D PARAM_MAX_PATH_DEPTH=7 -D PARAM_MAX_PATH_DEPTH_DIFFUSE=5 -D PARAM_MAX_PATH_DEPTH_GLOSSY=5 -D PARAM_MAX_PATH_DEPTH_SPECULAR=6 -D PARAM_RR_DEPTH=3 -D PARAM_RR_CAP=5.000000e-01f -D PARAM_SQRT_VARIANCE_CLAMP_MAX_VALUE=0.000000e+00f -D PARAM_IMAGE_FILTER_TYPE=5 -D PARAM_IMAGE_FILTER_WIDTH_X=1.500000e+00f -D PARAM_IMAGE_FILTER_WIDTH_Y=1.500000e+00f -D PARAM_IMAGE_FILTER_PIXEL_WIDTH_X=1 -D PARAM_IMAGE_FILTER_PIXEL_WIDTH_Y=1 -D PARAM_SAMPLER_TYPE=2 -D PARAM_SAMPLER_SOBOL_STARTOFFSET=32 -D LUXCORE_GENERIC_OPENCL
[LuxCore][10.907] [PathOCLBaseRenderThread::0] Compiling kernels 
ASSERTION FAILED: (isa<AllocaInst>(ptr) || ptrCandidate.empty()) && "storing/loading pointers only support private array"
  at file /build/beignet-jdrO6z/beignet-1.3.2/backend/src/llvm/llvm_gen_backend.cpp, function void gbe::GenWriter::findPointerEscape(llvm::Value*, std::set<llvm::Value*>&, bool, std::vector<llvm::Value*, std::allocator<llvm::Value*> >&), line 1004
Rilevato trace/breakpoint


Post Reply