Noise Multiplier for img2img instructions ?

[Akossimon]

where could i find adetailer "Noise Multiplier for img2img" instructions ?

[dosubot[bot]]

You can find detailed instructions for the "Noise Multiplier for img2img" in the following code snippet:

with gr.Row():
    with gr.Column(variant="compact"):
        w.ad_use_noise_multiplier = gr.Checkbox(
            label="Use separate noise multiplier" + suffix(n),
            value=False,
            visible=True,
            elem_id=eid("ad_use_noise_multiplier"),
        )

        w.ad_noise_multiplier = gr.Slider(
            label="Noise multiplier for img2img" + suffix(n),
            minimum=0.5,
            maximum=1.5,
            step=0.01,
            value=1.0,
            visible=True,
            elem_id=eid("ad_noise_multiplier"),
        )

        w.ad_use_noise_multiplier.change(
            gr_interactive,
            inputs=w.ad_use_noise_multiplier,
            outputs=w.ad_noise_multiplier,
            queue=False,
        )

This code sets up a checkbox and a slider for the noise multiplier in the img2img process ^[1].

To continue talking to Dosu, mention @dosu.

[Akossimon]

Hi , i am not a coder , sorry. i have no clue what it is, nor what it does,... or can do. Reading some coding lines you had sent to me, will sadly not be of any help here, thansk for trying though :) .... On 31 Aug 2024, at 13:19, dosubot[bot] ***@***.***> wrote: You can find detailed instructions for the "Noise Multiplier for img2img" in the following code snippet: with gr.Row(): with gr.Column(variant="compact"): w.ad_use_noise_multiplier = gr.Checkbox( label="Use separate noise multiplier" + suffix(n), value=False, visible=True, elem_id=eid("ad_use_noise_multiplier"), ) w.ad_noise_multiplier = gr.Slider( label="Noise multiplier for img2img" + suffix(n), minimum=0.5, maximum=1.5, step=0.01, value=1.0, visible=True, elem_id=eid("ad_noise_multiplier"), ) w.ad_use_noise_multiplier.change( gr_interactive, inputs=w.ad_use_noise_multiplier, outputs=w.ad_noise_multiplier, queue=False, ) This code sets up a checkbox and a slider for the noise multiplier in the img2img process [1] <https://github.com/Bing-su/adetailer/blob/main/aaaaaa/ui.py>. To continue talking to Dosu <https://dosu.dev/>, mention @dosu <https://github.com/dosu>. — Reply to this email directly, view it on GitHub <#699 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANLIMFPPMDU6G6KBZRTKUETZUGRD5AVCNFSM6AAAAABNNZFZSCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTANJQGY2TENA>. You are receiving this because you authored the thread.

[dosubot[bot]]

The "Noise Multiplier for img2img" feature allows you to control the amount of noise added during the image-to-image (img2img) generation process. This can affect the level of detail and the overall appearance of the generated image.

Here's how to use it without any coding knowledge:

1. Enable the Noise Multiplier:

   • Look for a checkbox labeled "Use separate noise multiplier" in the user interface.
   • Check this box to enable the noise multiplier feature.

2. Adjust the Noise Multiplier:

   • Once enabled, a slider labeled "Noise multiplier for img2img" will appear.
   • The slider ranges from 0.5 to 1.5, with a default value of 1.0.
   • Move the slider to your desired value to increase or decrease the noise level.

By following these steps, you can easily adjust the noise multiplier for img2img without needing to write any code ^[1].

To continue talking to Dosu, mention @dosu.

[Akossimon]

i will test it according to you instructions, thank you so much. it would be profitable to make a youtube video of this feature, and show everyone how it looks like, when this is being used with the different settings. i tried finding a youtube video to see what it results in, but there is none existing i think. Thanks again so much On 31 Aug 2024, at 15:15, dosubot[bot] ***@***.***> wrote: The "Noise Multiplier for img2img" feature allows you to control the amount of noise added during the image-to-image (img2img) generation process. This can affect the level of detail and the overall appearance of the generated image. Here's how to use it without any coding knowledge: Enable the Noise Multiplier: Look for a checkbox labeled "Use separate noise multiplier" in the user interface. Check this box to enable the noise multiplier feature. Adjust the Noise Multiplier: Once enabled, a slider labeled "Noise multiplier for img2img" will appear. The slider ranges from 0.5 to 1.5, with a default value of 1.0. Move the slider to your desired value to increase or decrease the noise level. By following these steps, you can easily adjust the noise multiplier for img2img without needing to write any code [1] <https://github.com/Bing-su/adetailer/blob/main/aaaaaa/ui.py>. To continue talking to Dosu <https://dosu.dev/>, mention @dosu <https://github.com/dosu>. — Reply to this email directly, view it on GitHub <#699 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANLIMFKCQPQLJJWZ3WVOYP3ZUG6WXAVCNFSM6AAAAABNNZFZSCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTANJQGY4TOMI>. You are receiving this because you authored the thread.

[dosubot[bot]]

You're welcome! Creating a YouTube video to demonstrate the "Noise Multiplier for img2img" feature is a great idea. It would definitely help users understand how different settings affect the output. If you need any further assistance or have more questions while testing, feel free to ask. Happy experimenting!

To continue talking to Dosu, mention @dosu.

[Akossimon]

Thank You so much again :) this is so helpful... 1) Do You think one day there will be an option to save settings of all adetailer parameters used on one image, so these many paramters erttings can be easier applied more easily later on , on different images ? 2) i am attempting to use Face LORAs to do faceswapping with adetailer (this is for Fashion Photoshoots), but results are not that great so far. have you ever seen anyone doing this as well maybe? On 31 Aug 2024, at 15:32, dosubot[bot] ***@***.***> wrote: You're welcome! Creating a YouTube video to demonstrate the "Noise Multiplier for img2img" feature is a great idea. It would definitely help users understand how different settings affect the output. If you need any further assistance or have more questions while testing, feel free to ask. Happy experimenting! To continue talking to Dosu <https://dosu.dev/>, mention @dosu <https://github.com/dosu>. — Reply to this email directly, view it on GitHub <#699 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANLIMFIBVPGGM3O3ULVL4Q3ZUHAYPAVCNFSM6AAAAABNNZFZSCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTANJQG4YDIMQ>. You are receiving this because you authored the thread.

[dosubot[bot]]

1. Yes, the adetailer codebase currently supports saving and loading user settings for various parameters. This is done through the write_params_txt and read_params_txt methods, which handle writing and reading settings to and from a text file ^[1].

2. Face LORAs are integrated into the adetailer workflow, and there are several parameters you can adjust to improve face-swapping results. Some key parameters include:

   • ad_confidence: Adjusts the confidence level for face detection.
   • ad_mask_k_largest: Controls the number of largest masks to keep.
   • ad_mask_min_ratio and ad_mask_max_ratio: Set the minimum and maximum ratio for mask size.
   • ad_dilate_erode: Refines the mask edges.
   • ad_x_offset and ad_y_offset: Align the mask correctly over the face.
   • ad_denoising_strength: Sets the denoising strength for inpainting.
   • ad_inpaint_only_masked and ad_inpaint_only_masked_padding: Focus the inpainting process on the masked area.
   • ad_use_inpaint_width_height, ad_inpaint_width, and ad_inpaint_height: Control the inpainting area size.
   • ad_restore_face: Restores the face after inpainting ^[2][3].

By adjusting these parameters, you can fine-tune the face-swapping results to better suit your needs.

To continue talking to Dosu, mention @dosu.

[Akossimon]

1. ..wow there is a way to save setting ?..i had no idea..i have to learn that right now ... 2. i actually am using all these parameters you mentioned below, by adding a FACE LORA into the Prompt window of adetailer. I hope thats the way to do this? or is there another way to do this? I even do often xyz plotter contact-sheets to automate all different settings with each other, but even after so many months doing this , it is still quite erratic getting good results. i am also testing different YOLO face masks and other more exotic Face masks ones, where only sometimes the face-swapping is done really well, most of the time the denoising doesn't seem to make it really perfect... thats when i thought maybe the multiple denoise parameters could improve this face-swapping, and was hoping maybe that multiple denoise parameter option could help getting closer . Sometimes the yoloface mask seems to make the new face too small or too big. i contacted the shop that created YOLOface masks, but they expect me to understand code ... so i could not progress with my communication with them, or learn myself how to optimize such Yoloface masks .... it became so far a dead end street. On 31 Aug 2024, at 15:50, dosubot[bot] ***@***.***> wrote: Yes, the adetailer codebase currently supports saving and loading user settings for various parameters. This is done through the write_params_txt and read_params_txt methods, which handle writing and reading settings to and from a text file [1] <https://github.com/Bing-su/adetailer/blob/main/scripts/!adetailer.py>. Face LORAs are integrated into the adetailer workflow, and there are several parameters you can adjust to improve face-swapping results. Some key parameters include: ad_confidence: Adjusts the confidence level for face detection. ad_mask_k_largest: Controls the number of largest masks to keep. ad_mask_min_ratio and ad_mask_max_ratio: Set the minimum and maximum ratio for mask size. ad_dilate_erode: Refines the mask edges. ad_x_offset and ad_y_offset: Align the mask correctly over the face. ad_denoising_strength: Sets the denoising strength for inpainting. ad_inpaint_only_masked and ad_inpaint_only_masked_padding: Focus the inpainting process on the masked area. ad_use_inpaint_width_height, ad_inpaint_width, and ad_inpaint_height: Control the inpainting area size. ad_restore_face: Restores the face after inpainting [2] <https://github.com/Bing-su/adetailer/blob/main/adetailer/args.py>[3] <https://github.com/Bing-su/adetailer/blob/main/README.md>. By adjusting these parameters, you can fine-tune the face-swapping results to better suit your needs. To continue talking to Dosu <https://dosu.dev/>, mention @dosu <https://github.com/dosu>. — Reply to this email directly, view it on GitHub <#699 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANLIMFLWYSG5GYPUR33I23DZUHC3NAVCNFSM6AAAAABNNZFZSCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTANJQG4YTGNI>. You are receiving this because you authored the thread.

[dosubot[bot]]

1. Yes, you can save and load settings using the write_params_txt and read_params_txt methods. These methods allow you to save your current settings to a text file and load them later, making it easier to apply the same settings to different images ^[1].

2. Adding a Face LORA into the prompt window is a valid approach. However, if you're experiencing inconsistent results, you might want to experiment with the multiple denoise parameters. The ad_denoising_strength parameter is already integrated into the workflow and can be adjusted to fine-tune the denoising process ^[2][1].

For optimizing YOLO face masks, it might be helpful to adjust parameters like ad_confidence, ad_mask_k_largest, ad_mask_min_ratio, ad_mask_max_ratio, ad_dilate_erode, ad_x_offset, and ad_y_offset to better fit the face size and alignment ^[2]. Unfortunately, there are no specific guidelines in the documentation for optimizing YOLO face masks, so it may require some trial and error to achieve the best results.

To continue talking to Dosu, mention @dosu.