c4diffusion.mesutcapkin.com

C4Diffusion v1. Docs.

—

von

c4diffusion_admin
in

C4DIFFUSION V1 – WORKS WITH:
a1111 sd webui at current state (date: 21.10.2023)

  • WHAT:
  • WHERE:
    • go to ⁠requirements and get the latest version
    • ongoing development for c4d R23+
    • supports c4d R20+, but frozen development at version v0_251 right now. will bring it up-to-date occasionally. testing period of the download has probably ended already by the time you read this. should you need a c4d r20+ version, send a message and i’ll update the link.
    • see ⁠most-recent-version for… well… just that
    • see ⁠bugfixes-and-changes for… well… just that
    • post bugs you find in ⁠bugs
  • HOW TO INSTALL:
    • go to ⁠how-to-installation-first-steps…
  • SUPPORTED EXTENSIONS, BUILT-IN AND EXTERNAL:
  • ADDITIONAL SUPPORT FOR:
  • THE PLUGIN ITSELF:
  • C4DIFFUSION OBJECTS:
    • define starting cmd-line args: minimum, custom, preset
    • define custom a1111 sd webui installation to use
    • utilize multiple gpu’s by running one instance of a1111’s sd webui server per gpu
    • define custom „output“-structure. in the plugin location, by default
  • C4DIFFUSION INTERFACE, TOP ROW:
    • c4diffusion object that is in use is shown
    • select gpu „on-the-fly“
    • select checkpoint and refiner-checkpoints per drop-down menu
    • define the weight of the refiner per slider.
  • C4DIFFUSION INTERFACE, 2nd ROW, TAB „Prompt / Neg.Prompt“:
    • text-field for prompt and negative prompt
    • weight slider for each of them
  • C4DIFFUSION INTERFACE, 2nd ROW, TAB „Clues for PromptGenerator“:
    • radio buttons: „None“, „Lexica.Art“, „Krea.Ai“, „GPT“
    • weight slider, that assigns the set value to the generated prompt
    • text-field: „clues“ for the selected generator to be fed with
    • „GPT SystemMsg txt“: use custom txt-file that contain „system“-messages for gpt, as the base for it to process your given „clues“ (one example-txt is provided in the plugin-folder)
    • „Browse Prompt Generator“: opens a dialog window, representing the chosen generator
  • C4DIFFUSION INTERFACE, 3rd ROW:
    • define x, y resolution
    • checkboxes: send the set resolutions to your rendersettings (or not)
  • C4DIFFUSION INTERFACE, 4th ROW:
    • set „Seed“.
    • random button to set „random“ seed. (represented as -1)
    • select „Sampler“ by drop-down menu
    • select „Resizing method“ by drop-down menu
    • define „Batch Size“, much faster generation of multiple images (with consecutive seeds) than one-by-one generation
    • define „Image Count“. generates set number of images (with consecutive seeds)
    • starting seed for both is the set seed (either custom or the one that is generated „randomly“)
    • example: „Image Count“ 3, „Batch Size“ 5, „Seed“ 0: total of 15 images are generated by generating (1. Image Count) 5 images starting with seeds 0 to 4, (2. Image Count) 5 images starting with seeds 5 to 9, (3. Image Count) 5 images starting with seeds 10 to 14
    • slider for „Sampling Steps“
    • slider for „Cfg Scale“
    • slider for „Denoising Strength“
  • C4DIFFUSION INTERFACE, 5th ROW, Buttons:
    • „TXT2IMG“: start rendering in txt2img-mode (triggers opening of SD-Viewer, if not already open, else: uses the open viewer)
    • „IMG2IMG“: start rendering in img2img-mode (triggers opening of SD-Viewer, if not already open, else: uses the open viewer)
    • „C4D-Render“: render a c4d-images
    • „Open SD-Viewer“: open the viewer representing rendered sd-images (if not already open)
    • „Browser WebUI“: open a1111’s sd webui interface in your web-browser
    • „SD to PV“: send your sd-renderings to the picture viewer
    • „Take Path from RS“: sets the paths defined in your rendersettings as the root for the „output“-structure. if none is given: saved project location will be set as the root. if project not saved: plugin-location becomes the root.
    • „Start SD“: start the (or additional) SD-server instance.
    • „Get Res. from Img“: take the resolution from the image currently visible in the „SD-Viewer“
  • C4DIFFUSION INTERFACE, 6th ROW, Checkboxes:
    • „Restore Faces“: select to activate „Restore Faces“ feature (a1111-feature)
    • „Tiling“: select to activate „Tiling“ feature (a1111-feature)
    • „Highres.Fix“: select to activate „Highres.Fix“ feature (a1111-feature)
    • „Upscale“: select to activate „Upscale“ feature (a1111-feature)
    • „Use Inpaint“: select to activate „Inpaint“ feature (a1111-feature)
    • if any of the following is checked, „TXT2IMG“ and/or „IMG2IMG“ will use the result as the init-image:
      • „1-Click“:
        • creates multiple rendersettings, based on your current rendersetting, leading to multiple c4d-renderings:
          • a rendering based on your original settings,
          • a rendering of a c4d depthmap-pass,
          • a rendering of a c4d obj-pass (1 per set object-channel, if present in your scene)
      • „Override Mats“:
        • creates additional rendersetting based on your current rendersetting – overrides your material settings of your rendering with a „clay“-like material
      • „HW-Render“:
        • creates additional rendersetting based on your current rendersetting – uses c4d hardware-rendering instaed of standard (or chosen) render engine
    • all of those three checkbox work in combination.
    • example: you have cube in your editor, have the checkbox „HW-Render“ activated and click „IMG2IMG“: a hardware rendering of your editor-view will be rendered and used as the init image, the sd-rendering process starts, the viewer opens up and you’ll witness the live preview in the viewer (if your „Live Update“ checkbox in the viewer is checked).
  • C4DIFFUSION INTERFACE, 7th ROW:
    • „Denoising Highres.Fix“, activated: uses the value of the slider for „Denoising Strength“ as the value for the Hires.Fix pass.
    • „Denoising Highres.Fix“, deactivated: no Denoising for Hires.Fix pass.
    • „Render into Material“: works with the standard c4d-Material that is put into the slot. sd-image rendering will be fed into that material.
  • C4DIFFUSION INTERFACE, 8th ROW, TABS DEPEND ON THE STATE OF YOUR SD (AND EXTENSIONS) INSTALLATION:
    • „ControlNet“-tabs:
      • checkbox „Enable ControlNet“: enable/disable this cn-instance.
      • drop-down menu: select „Balanced“, „My Prompt is more Important“, „ControlNet is more important“
      • checkbox „Low VRam“: activate that feature, if needed.
      • drop-down menu: resizing methods.
      • drop-down menu: preprocessor selection
      • drop-down menu: model selection
      • slider „Weight“ sets the weight for this cn-instance
      • „Load CN_Input“: use path to an image that you want to use as a cn (-preprocessor) input (instead of the c4d-rendering, chosen by methods described above)
      • tip: if you have a depth model selected, and set the preprocessor model to „none“ and you checked the checkbox „1-Click“, which renders also a c4d-depth map, this c4d-depth map will be used for the sd-rendering
      • „Img2Img Init-Custom“: use path to an image that you want to use as an input for img2img (in addition to (or instead of) the c4d-rendering, chosen by methods described above, if any)
      • slider „Guidance Start“: sets the weight for this cn-instance
      • slider „Guidance End“: sets the weight for this cn-instance
    • „Inpaint“-tab:
      • all options correspond with the same features within a1111’s sd-webui
    • „Extras“-tab:
      • two dropdown menus for to upscaler-models, corresponding with the feature variables in a1111’s sd-webui
    • „ExtraNetworks/Embeddings…“-tab:
      • „Use X“ checkbox uses the set attributes of the respective (X) feature
      • buttons open a dialog for the respective feature
      • see below for explanation of the respective dialogs
    • „ReActor Force“-tab:
      • see below for explanation
    • C4DIFFUSION INTERFACE, 9th ROW, Buttons:
      • „EXPERIMENT“: currently only there during development
      • „Toggle Mats“: activate/deactivate materials in c4d-editor
      • „Show Paths“: shows the paths of your chosen sd-installation and your „output“-structure
    • C4DIFFUSION DIALOG WINDOWS/VIEWERS/INTERFACES:
      • „Lora“, „Lycoris“, „Hypernet“, „Embedding“, VIEWER:
        • your folder structur defines the number of tabs and their respective naming.
        • checkbox activates/deactivate corresponding slider values (-5 to 5)
        • if activated: X will be added to the prompt, following the pattern of a1111’s emphasis-concept, e.g.: (yellow:value_of_the_slider) – no such thing, if slidervalue is 1.0
      • „Styles“, VIEWER:
        • same as the other viewers, but:
          • two checkboxes: left for positive and right for negative prompt.
          • two sliders: left for positive and right for negative prompt.
          • „Preview“-button: shows the preview image and the attributes (Title, Description, Prompt, Negative Prompt, Extra“) of X
          • „Generate“-button: uses the image currently in the viewer as the preivew image for this style
      • „ReActor“, TAB:
        • checkbox enable/disable
        • checkbox „Save Original“ – keeps the un-„ReActored“ version of the image
        • checkbox „Swap in source image“ – basically does nothing at this point
        • checkbox „Swap in generated image“ – swaps the face in your generated sd-renderings
        • „Path to Source Image“ – path to image which is has the face(s) that you want to incorporate into your sd-renderings
        • „Comma seperation“ for source and target: see explanation in tab itself
        • two drop-down menus for optional gender-detection
        • according to „HINT“: incorporate „Restore Faces“ and or „Upscale“ functionality as described.
  • „SD-Viewer“, VIEWER:
    • if not triggered by sd-rendering: opens in a size depending on the last sd-image you rendered.
    • if triggered by sd-rendering: opens in a size depending on your x, y settings
    • main area:
      • checkbox „Live Update“: acivate the live view of the generation process (needs the feature to be activated in your a1111’s settings in order to work)
      • slider: slide through your generated images
      • area shows the respective image / live preview
    • buttons area:
      • „Swtich CN“: if the sd-image you’re currently looking at has conrolnet-maps: switch back and forth to those and scrub through them (max. 3, due to the max. of controlnets instances supported by the plugin)
      • „Swtich Extras“: same concept, but for upscaled and „ReActored“ images
      • „Interrupt“: interrupt current sd-image generation
      • „Upscale“: upscale the image you’re currently looking at – upscaling-settings need to be set up in the „Extras“-tab
      • „Orig. Size“: snaps the view back to the original size of the image (if you’ve scaled view of the image with ctrl + mousewheel)
      • „Fetch Seed“: gets the seed of the current image and sets it in the „Seed“ of the main interface
      • „Show Infos“: opens a window (if not already open), showing: name of the generated file and parameters it had been generated with; also has a „Fetch Seed“-button
      • „Use as Init Image“: use the current image as the init image.
      • „Open Path“: open the explorer with the current image
      • „Send to Mat“: send current image to (c4d-standard-) matieral, that is set in the „Render into Material“-slot in main interface
      • „Send to PV“: sends current image to picture viewer
      • „ReActor This“: apply „ReActor“ to current image in viewer
  • „Browse Promptgenerator“, VIEWER, with Lexica.Art or Krea.Ai radio button set:
    • „Lexica.Art“ and „Krea.Ai“ follow the same concept.
    • functions similar to above viewers.
    • slider does NOT scrub through images, but through the „Generation Info“, resulting from your given „clues“. to see an image: click on „Show Image“ or „Show in Site“
    • „Use Prompt“: add the respective prompt to your prompt (not into your own prompt text-field but to the resulting „assembled total“ prompt
    • „To Clipboard“-button: copies the prompt in the „Generation Info“ to your clipboard.
    • „Refresh“: if you change your „clues“, while viewer is open, new results will be generated.
  • „Browse Promptgenerator“, VIEWER, GPT, with GPT radio button set:
    • need to have a working api-account at OpenAi.
    • if „GPT SystemMsg txt“ is set: results will be accordingly. otherwise, it’s „just“ in chat mode
    • rest: same as the other viewers.
    • drop-down menu lets you choose, which model you want to use
  • C4DIFFUSION AND ANIMATION:
    • if your original rendersettings have set a range, instead of a single image, sd-generation will be over that range.

Kommentare

Eine Antwort zu „C4Diffusion v1. Docs.“

  1. Avatar von A WordPress Commenter
    A WordPress Commenter

    Hi, this is a comment.
    To get started with moderating, editing, and deleting comments, please visit the Comments screen in the dashboard.
    Commenter avatars come from Gravatar.

    Antworten

Schreibe einen Kommentar zu A WordPress Commenter Antwort abbrechen

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert