jackMort/ChatGPT.nvim

[Feature request]: Add support for image generation/editing models

Open

#386 opened on 2024年2月4日

GitHub で見る
 (0 comments) (0 reactions) (0 assignees)Lua (4,005 stars) (323 forks)batch import
enhancementhelp wanted

説明

It would be really cool to support image generation and/or image editing models via this plugin. For example: dall-e-2 and dall-e-3 reuse the same OpenAI API key as regular GPT models. (docs).

Currently, when I manually set the model setting to dall-e-3 I get this response: image But I am successfully generating an image if I manually query OpenAI via curl image

Proposal:

  • By default the plugin's response to a prompt is to print the revised_prompt and the newly saved image path + filename.
  • By default the plugin saves all image responses to some location.
    • Default value for this download location is set via the config object.
    • Optionally this can be toggled off, and user will have to open URLs with gx to preview images.
  • When using dall-e-2/dall-e-3 the plugin can accept a new setting for the filesystem path where artifacts can be saved.
    • This way, you can use separate directories for different "visions" or projects

コントリビューターガイド

技術スタック
なし
領域
aideveloper experience
Issue 種別
feature
難度新しい貢献者にとっての実装難度です。1 は非常に小さな変更、5 は専門的な作業です。
3
推定時間経験ある貢献者が調査、実装、テスト、pull request 準備にかけるおおよその時間範囲です。
1-2 days
活動状況Issue が今どれだけ取り組みやすいかを示します。新しい、活発、古い、ブロック中、またはメンテナー入力待ちなどです。
fresh
明確さ期待される変更、受け入れ条件、次の手順がどれだけ明確かを示します。
clear
前提条件
Lua programmingOpenAI API basicsNeovim plugin concepts
初心者向け度初回貢献者にどれだけ取り組みやすいかを 1-100 で推定したスコアです。
50
調査方針
Investigate how the plugin currently handles API calls to OpenAI, likely in a file like `chatgpt.lua`. The issue proposes modifications to support image generation endpoints and handling file saves. Examine the plugin's configuration structure to add new options for image download path. Review the OpenAI API documentation for image generation to understand response format. Since there are no comments or PRs, start by familiarizing with the codebase and propose a design.