Frustrating couple of issues

[dm3rd]

I haven't updated in a few months or something, so when the max vram option got added I decided to update.

with a fresh download and settings file I got two issues.

when using the option to send gen info from the gallery it freezes up the ui and pops up a warning about slowing down etc.
sending info to txt2img and upscale seems fine,
sending to img2img and any2video causes the 'slow down'
sending to imgedit provides the below error
i'm on firefox and windows 10

Traceback (most recent call last):
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\gradio\queueing.py", line 867, in process_events
    response = await route_utils.call_process_api(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\Ben\AppData\Local\Programs\Python\Python311\Lib\contextlib.py", line 222, in __aexit__
    await self.gen.athrow(typ, value, traceback)
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\gradio\profiling.py", line 209, in trace_phase
    yield
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\gradio\route_utils.py", line 374, in call_process_api
    output = await app.get_blocks().process_api(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\Ben\AppData\Local\Programs\Python\Python311\Lib\contextlib.py", line 222, in __aexit__
    await self.gen.athrow(typ, value, traceback)
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\gradio\profiling.py", line 209, in trace_phase
    yield
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\gradio\blocks.py", line 2192, in process_api
    data = await self.postprocess_data(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\gradio\blocks.py", line 1955, in postprocess_data
    prediction_value = await anyio.to_thread.run_sync(
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\anyio\to_thread.py", line 63, in run_sync
    return await get_async_backend().run_sync_in_worker_thread(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\anyio\_backends\_asyncio.py", line 2518, in run_sync_in_worker_thread
    return await future
           ^^^^^^^^^^^^
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\anyio\_backends\_asyncio.py", line 1002, in run
    result = context.run(func, *args)
             ^^^^^^^^^^^^^^^^^^^^^^^^
  File "F:\sdcppwebui\sd.cpp-webui\venv\Lib\site-packages\gradio\components\gallery.py", line 251, in postprocess
    raise ValueError(
ValueError: The `value` passed into `gr.Gallery` must be a list of images or videos, or list of (media, caption) tuples.

another issue I came across the 'offload to cpu' options casues an instant gen fail.
but using the new max vram option mostly bypasses it. not sure if thats the ui or sd.cpp issue.
EDIT:: after some more testing seems the max vram isn't doing as much as I had hoped.

the only other issue i've come across is that the quantization setting is saved but doesn't get used when loading the ui.
I also have to set it to what I need as it's always default

[daniandtheweb]

This is quite a lot to unpack.

1. I'm currently investigating the UI freeze issue. I think it may be related to the recent Gradio update, I'll test it and, if it is, I'll just downgrade it.
2. I'll investigate the "send to imgedit issue".
3. I'll test the offload to cpu fail issue.
4. Max Vram seems to limit the VRAM buffers, you'll get actual VRAM savings if you use it together with offload to cpu.
5. Some settings haven't been behaving as they should for a while, like the quantization settings. Once I fix the other issues mentioned here I can take a look at it.

[daniandtheweb]

I think I managed to fix point the tab switching issue and the send to imgedit issue. You can try to update to check the new version.

I haven't been able to reproduce the offload to cpu issue, I guess that is related to sd-cli. Is there any log you can send in the command prompt?

[dm3rd]

All I get is this

sd-cli.exe -M img_gen -p "person, formal clothes, city park" -n "ugly, bad art" --sampling-method dpm++2s_a --steps 6 -W 832 -H 1024 -b 1 --cfg-scale 1.5 -s -1 --clip-skip 2 --embd-dir F:\sdcppwebui\sd.cpp-webui\models/embeddings/ --lora-model-dir F:\sdcppwebui\sd.cpp-webui\models/loras/ -t 0 --rng cpu --sampler-rng cpu --lora-apply-mode auto -o F:\sdcppwebui\sd.cpp-webui\outputs/txt2img/20260509_184635_078417_6_steps.png --model F:\sdcppwebui\sd.cpp-webui\models/checkpoints/prefectPonyXL_v50.safetensors --vae F:\sdcppwebui\sd.cpp-webui\models/vae/irisXLVAE_luna.safetensors --type q4_K --scheduler karras --max-vram 3.8 --vae-tile-overlap 0.5 --vae-tile-size 32x32 --offload-to-cpu --vae-tiling --fa --diffusion-conv-direct --vae-conv-direct --mmap --color

stable-diffusion.cpp:212 - loading model from 'F:\sdcppwebui\sd.cpp-webui\models/checkpoints/prefectPonyXL_v50.safetensors'
[INFO ] model.cpp:219 - load F:\sdcppwebui\sd.cpp-webui\models/checkpoints/prefectPonyXL_v50.safetensors using safetensors format
[INFO ] stable-diffusion.cpp:280 - loading vae from 'F:\sdcppwebui\sd.cpp-webui\models/vae/irisXLVAE_luna.safetensors'
[INFO ] model.cpp:222 - load F:\sdcppwebui\sd.cpp-webui\models/vae/irisXLVAE_luna.safetensors using torch zip format
[INFO ] stable-diffusion.cpp:305 - Version: SDXL
[INFO ] stable-diffusion.cpp:333 - Weight type stat: f16: 2513
[INFO ] stable-diffusion.cpp:334 - Conditioner weight type stat: f16: 585
[INFO ] stable-diffusion.cpp:335 - Diffusion model weight type stat: f16: 1680
[INFO ] stable-diffusion.cpp:336 - VAE weight type stat: f16: 248
[INFO ]
Subprocess terminated.

[daniandtheweb]

This looks like an issue with stable-diffusion.cpp, it only happens when using offload-to-cpu?

[dm3rd]

I used a few different option combos and only came up when offload-to-cpu was active.
when it was off it always generated. i'm gonna test it some more though

[dm3rd]

I think you might be right, I tested it outside of the ui and got the same issue.
gonna test with some older exe files