I am currently in the process of building a front end back end interface that allows me to use your extension to queue up jobs from a website. A lot of it I've got working. But I'm really struggling to get a detailer to work with the /agent-scheduler/v1/queue/img2img api.
The reference shows that I should be using "alwayson_scripts": {}, and currently that's what I'm using when doing normal image Generations, and it works fine using it that way. But when I try to use it under the API reference above, it just doesn't engage ADetailer at all. When I manually set up a job it automatic 1111 under the in-paint tab in img2img, and enqueue the job the payload that it sends looks like this:
as you can see its using extra_generation_params instead of "alwayson_scripts": {},
its not giving me any errors, the images do have a face/body. but it just does not seem to engage with it when i try either way.
here is the payload i have built up:
{
"prompt: "portrait of a woman",
"width": 1024,
"height": 1024,
"sampler_name": "DPM++ 2M SDE Turbo",
"cfg_scale": 3,
"alwayson_scripts": {
"ADetailer": {
"args": [
true,
true,
{
"ad_model": "face_yolov8n.pt",
"ad_checkpoint": "dreamshaperXL_lightningInpaint.safetensors",
"ad_use_checkpoint": true,
"ad_cfg_scale": 2.5,
"ad_denoising_strength": 0.4,
"ad_inpaint_width": 1024,
"ad_inpaint_height": 1024,
"ad_mask_blur": 8,
"ad_dilate_erode": 16,
"ad_use_inpaint_width_height": true
},
false
]
}
},
"steps": 12,
"checkpoint": "dreamshaperXL_lightningInpaint.safetensors",
"denoising_strength": 1,
"is_using_inpainting_conditioning": true,
"inpainting_fill": 0,
"inpainting_mask_invert": 0,
"include_init_images": true,
"mask_blur_x": 12,
"mask_blur_y": 12
}
When i am just generating a image using the txt2img api, the adetailer works fine, see my jason package below:
Payload: {
"prompt": "Portrait painting, (Regal figure in flowing robes:1.3), Balanced composition, Crowned head, Commanding presence, (Rich jewel tones:1.2), Intricate details, Majestic backdrop, Royal air, (Soft diffused light:1.3)",
"cfg_scale": "4",
"steps": "8",
"height": 1024,
"width": 1024,
"seed": "-1",
"checkpoint": "SDXL-Turbo-LCM\RealitiesEdgeXLLIGHTNING_TURBOV7.safetensors",
"alwayson_scripts": {
"ADetailer": {
"args": [
true,
true,
{
"ad_model": "face_yolov8n.pt",
"ad_checkpoint": "dreamshaperXL_lightningInpaint.safetensors",
"ad_use_checkpoint": true,
"ad_cfg_scale": 2.5,
"ad_denoising_strength": 0.4,
"ad_inpaint_width": 1024,
"ad_inpaint_height": 1024,
"ad_mask_blur": 8,
"ad_dilate_erode": 16,
"ad_use_inpaint_width_height": true
},
false
]
}
},
"enable_hr": false,
"hr_upscaler": "4x_NMKD-Siax_200k",
"hr_second_pass_steps": 10,
"denoising_strength": 0.2,
"hr_scale": 2,
"hr_resize_x": 0,
"hr_resize_y": 0,
"hr_checkpoint_name": "Use same checkpoint",
"hr_prompt": "",
"hr_negative_prompt": "",
"sampler_name": "DPM++ 2M SDE Turbo"
}
I guess what i would like to know is there someone who can give me a good example of what the json package should look like for this when using the img2img api model to do inpainting and what the adetailer would look like.
I am currently in the process of building a front end back end interface that allows me to use your extension to queue up jobs from a website. A lot of it I've got working. But I'm really struggling to get a detailer to work with the /agent-scheduler/v1/queue/img2img api. The reference shows that I should be using "alwayson_scripts": {}, and currently that's what I'm using when doing normal image Generations, and it works fine using it that way. But when I try to use it under the API reference above, it just doesn't engage ADetailer at all. When I manually set up a job it automatic 1111 under the in-paint tab in img2img, and enqueue the job the payload that it sends looks like this:
"prompt": "portrait of a woman", "all_prompts": ["portrait of a woman"], "negative_prompt": "", "all_negative_prompts": [""], "seed": 1800119224, "all_seeds": [1800119224], "subseed": 1047218332, "all_subseeds": [1047218332], "subseed_strength": 0, "width": 1024, "height": 1024, "sampler_name": "DPM++ 2M SDE Turbo", "cfg_scale": 3, "steps": 13, "batch_size": 1, "restore_faces": false, "face_restoration_model": null, "sd_model_name": "dreamshaperXL_lightningInpaint", "sd_model_hash": "1a49cd4473", "sd_vae_name": null, "sd_vae_hash": null, "seed_resize_from_w": -1, "seed_resize_from_h": -1, "denoising_strength": 1, "extra_generation_params": {"ADetailer model": "face_yolov8n.pt", "ADetailer confidence": 0.3, "ADetailer dilate erode": 16, "ADetailer mask blur": 8, "ADetailer denoising strength": 0.4, "ADetailer inpaint only masked": true, "ADetailer inpaint padding": 32, "ADetailer use separate steps": true, "ADetailer steps": 6, "ADetailer use separate checkpoint": true, "ADetailer checkpoint": "dreamshaperXL_lightningInpaint.safetensors [1a49cd4473]", "ADetailer use separate sampler": true, "ADetailer sampler": "DPM++ 2M SDE Turbo", "ADetailer model 2nd": "person_yolov8n-seg.pt", "ADetailer confidence 2nd": 0.3, "ADetailer dilate erode 2nd": 4, "ADetailer mask blur 2nd": 8, "ADetailer denoising strength 2nd": 0.4, "ADetailer inpaint only masked 2nd": true, "ADetailer inpaint padding 2nd": 32, "ADetailer version": "24.5.1", "Denoising strength": 1, "Mask blur": 12, "Inpaint area": "Only masked", "Masked area padding": 56, "Masked content": "fill"}, "index_of_first_image": 0, }
as you can see its using extra_generation_params instead of "alwayson_scripts": {},
its not giving me any errors, the images do have a face/body. but it just does not seem to engage with it when i try either way.
here is the payload i have built up: { "prompt: "portrait of a woman", "width": 1024, "height": 1024, "sampler_name": "DPM++ 2M SDE Turbo", "cfg_scale": 3, "alwayson_scripts": { "ADetailer": { "args": [ true, true, { "ad_model": "face_yolov8n.pt", "ad_checkpoint": "dreamshaperXL_lightningInpaint.safetensors", "ad_use_checkpoint": true, "ad_cfg_scale": 2.5, "ad_denoising_strength": 0.4, "ad_inpaint_width": 1024, "ad_inpaint_height": 1024, "ad_mask_blur": 8, "ad_dilate_erode": 16, "ad_use_inpaint_width_height": true }, false ] } }, "steps": 12, "checkpoint": "dreamshaperXL_lightningInpaint.safetensors", "denoising_strength": 1, "is_using_inpainting_conditioning": true, "inpainting_fill": 0, "inpainting_mask_invert": 0, "include_init_images": true, "mask_blur_x": 12, "mask_blur_y": 12 }
When i am just generating a image using the txt2img api, the adetailer works fine, see my jason package below: Payload: { "prompt": "Portrait painting, (Regal figure in flowing robes:1.3), Balanced composition, Crowned head, Commanding presence, (Rich jewel tones:1.2), Intricate details, Majestic backdrop, Royal air, (Soft diffused light:1.3)", "cfg_scale": "4", "steps": "8", "height": 1024, "width": 1024, "seed": "-1", "checkpoint": "SDXL-Turbo-LCM\RealitiesEdgeXLLIGHTNING_TURBOV7.safetensors", "alwayson_scripts": { "ADetailer": { "args": [ true, true, { "ad_model": "face_yolov8n.pt", "ad_checkpoint": "dreamshaperXL_lightningInpaint.safetensors", "ad_use_checkpoint": true, "ad_cfg_scale": 2.5, "ad_denoising_strength": 0.4, "ad_inpaint_width": 1024, "ad_inpaint_height": 1024, "ad_mask_blur": 8, "ad_dilate_erode": 16, "ad_use_inpaint_width_height": true }, false ] } }, "enable_hr": false, "hr_upscaler": "4x_NMKD-Siax_200k", "hr_second_pass_steps": 10, "denoising_strength": 0.2, "hr_scale": 2, "hr_resize_x": 0, "hr_resize_y": 0, "hr_checkpoint_name": "Use same checkpoint", "hr_prompt": "", "hr_negative_prompt": "", "sampler_name": "DPM++ 2M SDE Turbo" } I guess what i would like to know is there someone who can give me a good example of what the json package should look like for this when using the img2img api model to do inpainting and what the adetailer would look like.