Hi,
I'd love for image editing capabilities to be added.
User Story
Ideally, I'd like to be able to have a chat and ask for edits to be made to the previous or original image with capabilities/behaviors similar to regular chat applications.
Current Limitation
I believe edits are currently done by regenerating the image, as attachments appear to not be passed through the image generating tool.
|
image = try await generateImageUseCase.execute(prompt: prompt, model: modelId, attachments: []) |
Use Case
From a UX perspective, the application currently requires the user to save and re-attach each image. It would be nice should:
- Previous images be attached automatically to the conversation or long-press actions on chat images allow for individual image re-attachment.
- Support pasting images directly into the chat box (e.g., copy image from a web browser and paste it in the chat for an edit).
Furthermore, it would be nice should attached images show up as miniatures or in-lined within the prompt, enabling the user to ask edits to image A (generation/edit tool) while referring to content of image B (vision). On iOS, names of attached images from the Photo Library are not intuitive making it hard to distinguish between attachments. Similarly, attaching images from the Photo Library should allow multiple images to be selected instead of requiring the user to perform the attachment steps repeatedly.
I hope you'll take such features into consideration,
Thanks for the pleasant application!
Hi,
I'd love for image editing capabilities to be added.
User Story
Ideally, I'd like to be able to have a chat and ask for edits to be made to the previous or original image with capabilities/behaviors similar to regular chat applications.
Current Limitation
I believe edits are currently done by regenerating the image, as attachments appear to not be passed through the image generating tool.
openclient-llm/openclient-llm/Shared/Features/Chat/Tools/GenerateImageTool.swift
Line 82 in 977d502
Use Case
From a UX perspective, the application currently requires the user to save and re-attach each image. It would be nice should:
Furthermore, it would be nice should attached images show up as miniatures or in-lined within the prompt, enabling the user to ask edits to image A (generation/edit tool) while referring to content of image B (vision). On iOS, names of attached images from the Photo Library are not intuitive making it hard to distinguish between attachments. Similarly, attaching images from the Photo Library should allow multiple images to be selected instead of requiring the user to perform the attachment steps repeatedly.
I hope you'll take such features into consideration,
Thanks for the pleasant application!