(Feature Request) Apply multi loras into regional prompt and upscale with new extension Lora Mark

[wardensc2]

Hi pkuliyi2015, someone name lifeisboringsoprogramming just make a loramark extension which help us to apply multi loras into any region we want and also can use hires fix to upscale, I try to use it with regional prompt and upscale but so far it not work.

Either we lose region prompt or multi loras, we have to choose. I really love region prompt to draw good background incorporate with characters but I also want to upgrade regional prompt more with new Lora Mark extension can you support us

Thank in advance

Here is the link of the new extension Lora Mark

https://github.com/lifeisboringsoprogramming/sd-webui-lora-masks

[Vendaciousness]

I had the same problem trying to use ControlNet with Region Control, in both text2image and image2image. I'd write up a bug, but I already wasted the night, thinking it would work and getting garbage. Also, the regions were incorrect. I put the bounding boxes over the faces of this image with correct WxH and got this:

Here's what I had for regions:

[wardensc2]

I had the same problem trying to use ControlNet with Region Control, in both text2image and image2image. I'd write up a bug, but I already wasted the night, thinking it would work and getting garbage. Also, the regions were incorrect. I put the bounding boxes over the faces of this image with correct WxH and got this:

Here's what I had for regions:

Do your controlnet image position as same as the box of the region control, I never experience this.

[Vendaciousness]

You know what? I think it may have been "Enable Dynamic Thresholding" that was causing the problems, but I have to test more to be sure. Here's the Traceback I get when I have it enabled:

MultiDiffusion Sampling:  42%|██████▋         | 75/180 [01:57<02:44,  1.57s/it]
[Tiled Diffusion] ControlNet found, support is enabled.
23:15:31-521591 INFO     Loading model from cache: control_v11p_sd15_openpose
                         [cab727d4]
23:15:31-525591 INFO     Loading preprocessor: openpose_full
23:15:31-527591 INFO     preprocessor resolution = 640
23:15:32-509591 INFO     Loading model from cache: control_v11p_sd15_canny
                         [d14c016b]
23:15:32-514595 INFO     Loading preprocessor: canny
23:15:32-516591 INFO     preprocessor resolution = 640
23:15:33-538591 INFO     Loading model from cache: control_v11p_sd15_canny
                         [d14c016b]
23:15:33-542591 INFO     Loading preprocessor: canny
23:15:33-545592 INFO     preprocessor resolution = 640
23:15:34-620591 INFO     Loading model from cache: control_v11f1e_sd15_tile
                         [a371b31b]
23:15:34-625592 INFO     Loading preprocessor: tile_resample
23:15:34-627591 INFO     preprocessor resolution = 640
MultiDiffusion Sampling:   0%|                         | 0/180 [00:00<?, ?it/s]MultiDiffusion hooked into 'DPM++ 2M Karras' sampler, Tile size: 96x96, Tile batches: 0, Batch size: 4. (ext: RegionCtrl, ContrlNet)
[Tiled VAE]: the input size is tiny and unnecessary to tile.
23:15:36-008591 ERROR    Running script process batch:
                         G:\Code\automatic\extensions-builtin\sd-dynamic-thresh
                         olding\scripts\dynamic_thresholding.py: AttributeError
┌───────────────────── Traceback (most recent call last) ─────────────────────┐
│ G:\Code\automatic\modules\scripts.py:430 in process_batch                   │
│                                                                             │
│   429 │   │   │   │   args = p.per_script_args.get(script.title(), p.script │
│ > 430 │   │   │   │   script.process_batch(p, *args, **kwargs)              │
│   431 │   │   │   │   s.append(f'{script.title()}:{round(time.time()-t0, 2) │
│                                                                             │
│ G:\Code\automatic\extensions-builtin\sd-dynamic-thresholding\scripts\dynami │
│ c_thresholding.py:128 in process_batch                                      │
│                                                                             │
│   127 │   │   if p.sampler is not None:                                     │
│ > 128 │   │   │   p.sampler = sd_samplers.create_sampler(fixed_sampler_name │
│   129                                                                       │
│                                                                             │
│ G:\Code\automatic\extensions-builtin\multidiffusion-upscaler-for-automatic1 │
│ 111\scripts\tilediffusion.py:362 in <lambda>                                │
│                                                                             │
│   361 │   │   sd_samplers.create_sampler_original_md = sd_samplers.create_s │
│ > 362 │   │   sd_samplers.create_sampler = lambda name, model: self.create_ │
│   363 │   │   │   name, model, p, Method(method),                           │
│                                                                             │
│ G:\Code\automatic\extensions-builtin\multidiffusion-upscaler-for-automatic1 │
│ 111\scripts\tilediffusion.py:434 in create_sampler_hijack                   │
│                                                                             │
│   433 │   │   # create a sampler with the original function                 │
│ > 434 │   │   sampler = sd_samplers.create_sampler_original_md(name, model) │
│   435 │   │   if method == Method.MULTI_DIFF: delegate_cls = MultiDiffusion │
└─────────────────────────────────────────────────────────────────────────────┘
AttributeError: module 'modules.sd_samplers' has no attribute
'create_sampler_original_md'

Sorry, I didn't realize I had it on. I'll retest and write up a bug if there's still an issue without it.