The Power Dall-E local web UI is a simple way to use Dall-E 3 with its advanced parameters like style (vivid vs natural) or quality (standard vs hd), which aren't available via ChatGPT. You can also use it to keep generating images when your main ChatGPT Dall-E is throttling you. This app locally remembers your past creations and shows them in a newest-first list, including your prompt as well as the prompt Dall-E rewrote behind the scenes.
With Power Dall-E, you can launch multiple generations at once in any number, which makes it faster than ChatGPT and better to A/B test prompts, if you are willing to carry the extra API cost. One other difference to ChatGPT is that Power Dall-E doesn't remember any conversation session, something which can lead ChatGPT (sometimes for better, sometimes for worse) to bias successive prompt rewrites to the ongoing conversation.
This app does not connect to any server outside of your local one and OpenAI. Be aware of the costs though, as every API call is paid to OpenAI as per their usual pricing (note I'm not part of that transaction nor do I make money from it, it's just a local way for you to connect via your OpenAI API key).
As you scroll down, your older creations also show.
- Ensure you have Node.js installed.
- Load this git project onto your local drive.
- Create a name-less file ".env" in the project root and put into it your OpenAI API key by writing: OPENAI_API_KEY=yourapikey
- Navigate to the project via the command line, and type npm install
- Then in the same command line, type npm start
- Open http:https://localhost:9000 in your browser. Have fun!
This project makes no guarantees for being bug-free, use at your own risk, and keep in mind that API calls cost money (see the OpenAI costs dashboard and the pricing table).
Power Dall-E is integrated with Prompt Inspirer (also available at PromptInspirer.com), which includes over 10,000 items you can drag & drop together to make prompts.
Prompt Inspirer overview video
Info such as the prompt used are stored in a local database. If you want to additionally save a JSON file with such info with each image, add this to your ".env" file: SAVE_JSON_WITH_IMAGES=true
To switch from default model "dall-e-3" to another, add e.g. this to the ".env" file and restart the server (though size dropdowns and such will only properly work for the default model): MODEL=dall-e-2
To change the maximum number of images served at the start (the default is 1000), use e.g.: MAX_IMAGES_TO_SERVE_AT_START=2000