mirror of
https://github.com/discourse/discourse-ai.git
synced 2025-06-30 19:42:17 +00:00
17f04c76d8
This commit enhances the AI image generation functionality by adding support for: 1. OpenAI's GPT-based image generation model (gpt-image-1) 2. Image editing capabilities through the OpenAI API 3. A new "Designer" persona specialized in image generation and editing 4. Two new AI tools: CreateImage and EditImage Technical changes include: - Renaming `ai_openai_dall_e_3_url` to `ai_openai_image_generation_url` with a migration - Adding `ai_openai_image_edit_url` setting for the image edit API endpoint - Refactoring image generation code to handle both DALL-E and the newer GPT models - Supporting multipart/form-data for image editing requests * wild guess but maybe quantization is breaking the test sometimes this increases distance * Update lib/personas/designer.rb Co-authored-by: Alan Guo Xiang Tan <gxtan1990@gmail.com> * simplify and de-flake code * fix, in chat we need enough context so we know exactly what uploads a user uploaded. * Update lib/personas/tools/edit_image.rb Co-authored-by: Alan Guo Xiang Tan <gxtan1990@gmail.com> * cleanup downloaded files right away * fix implementation --------- Co-authored-by: Alan Guo Xiang Tan <gxtan1990@gmail.com>
78 lines
2.3 KiB
Ruby
78 lines
2.3 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
module DiscourseAi
|
|
module AiHelper
|
|
class Painter
|
|
def commission_thumbnails(input, user)
|
|
return [] if input.blank?
|
|
|
|
model = SiteSetting.ai_helper_illustrate_post_model
|
|
|
|
if model == "stable_diffusion_xl"
|
|
stable_diffusion_prompt = diffusion_prompt(input, user)
|
|
return [] if stable_diffusion_prompt.blank?
|
|
|
|
artifacts =
|
|
DiscourseAi::Inference::StabilityGenerator
|
|
.perform!(stable_diffusion_prompt)
|
|
.dig(:artifacts)
|
|
.to_a
|
|
.map { |art| art[:base64] }
|
|
|
|
base64_to_image(artifacts, user.id)
|
|
elsif model == "dall_e_3"
|
|
attribution =
|
|
I18n.t(
|
|
"discourse_ai.ai_helper.painter.attribution.#{SiteSetting.ai_helper_illustrate_post_model}",
|
|
)
|
|
results =
|
|
DiscourseAi::Inference::OpenAiImageGenerator.create_uploads!(
|
|
input,
|
|
model: "dall-e-3",
|
|
user_id: user.id,
|
|
title: attribution,
|
|
)
|
|
results.map { |result| UploadSerializer.new(result[:upload], root: false) }
|
|
end
|
|
end
|
|
|
|
private
|
|
|
|
def base64_to_image(artifacts, user_id)
|
|
attribution =
|
|
I18n.t(
|
|
"discourse_ai.ai_helper.painter.attribution.#{SiteSetting.ai_helper_illustrate_post_model}",
|
|
)
|
|
|
|
artifacts.each_with_index.map do |art, i|
|
|
f = Tempfile.new("v1_txt2img_#{i}.png")
|
|
f.binmode
|
|
f.write(Base64.decode64(art))
|
|
f.rewind
|
|
upload = UploadCreator.new(f, attribution).create_for(user_id)
|
|
f.unlink
|
|
|
|
UploadSerializer.new(upload, root: false)
|
|
end
|
|
end
|
|
|
|
def diffusion_prompt(text, user)
|
|
prompt =
|
|
DiscourseAi::Completions::Prompt.new(
|
|
<<~TEXT.strip,
|
|
Provide me a StableDiffusion prompt to generate an image that illustrates the following post in 40 words or less, be creative.
|
|
You'll find the post between <input></input> XML tags.
|
|
TEXT
|
|
messages: [{ type: :user, content: text, id: user.username }],
|
|
)
|
|
|
|
DiscourseAi::Completions::Llm.proxy(SiteSetting.ai_helper_model).generate(
|
|
prompt,
|
|
user: user,
|
|
feature_name: "illustrate_post",
|
|
)
|
|
end
|
|
end
|
|
end
|
|
end
|