-
Images generated with prepare_su_noise_multires don't appear to be deterministic.
I suspect interpolate as the source of issue.
`
noise = torch.nn.functional.interpolate(noise, size=latent_image.…
-
tried training my dataset in 3 different resolutions, trained for 5k steps, it learnt nothing.
-
# Tasks
- [ ] 1. implement blender export for _low _high mesh (objname+subdivlevel+_low/_high)
- [ ] 2. implement blender import for _low _high mesh (recognize type by suffix and reshape with mul…
-
As progressive image formats (like FLIF) aren't supported yet, it would increase performance and bandwith utilization significantly,
when multiresolution zooming is implemented. This means that the …
-
1.
set "noise_offset_type": "Multires",
set "multires_noise_discount": 0.8,
2.
See training failed (overtrainig). stop (or load setting)
4.
set "noise_offset_type": "Original",
5.
it …
-
Hi, I find part of the reconstructed mesh is missing. I guess tuning some parameters can be helpful. The following is a screenshot of the Bronze reconstruction (200k iteration). I use the default para…
-
My config files are attached. This example is of an architectural drawing pdf that was converted into images.
A few assumptions just to make sure I understand the file structure and config correctl…
-
I built an examplepage with more Pannellum-Scenes one from typ **equirectangular** and one from type **multires**. I used on both Scenes the same equirectangular-panorama-jpg. For the **multires** I c…
-
this is awesome project but visible calculator is too complex.
I suggestion calculator visible by Frustum
https://stackoverflow.com/questions/10858599/how-to-determine-if-plane-is-in-three-js-camera…
-
This extension is much more efficient and simple to use than kohya. I like it a lot!
However I am having a frequent issue where it will fill up the memory right before training. After the following…