Most AI tools are built around chat interfaces. That works for generating an occasional image, but it’s not very efficient for daily production work. AI Lab Standalone replaces the chat with a visual workflow:
This allows you to evaluate and compare image quality directly.
All generated images, references, and prompts are stored directly on your hard drive. Nothing is hidden in a cloud session, nothing tied to a browser.
AI rarely fails because of image quality, it fails because the result doesn’t match what you had in mind. AI Lab Standalone helps close this gap:
The goal is not just faster generation, but fewer iterations during generation.
This keeps full control over what is sent to the AI at all times.
Optionally, results can be sent directly to Photoshop, including very old Photoshop versions.
AI Lab is not a playground, it's your workspace for generative images.
The same tokens work in both the Standalone workspace and the Photoshop UXP plugin.
We use the official APIs from Google, ByteDance, Black Forest Labs and OpenAI and do not have any special agreements with them. Therefore, the standard API usage terms of the respective providers apply.
Google, OpenAI and ByteDance apparently do not use submitted images for training purposes:
Google (Nano Banana): https://ai.google.dev/gemini-api/docs/zdr?hl=en
OpenAI (GPT Image): https://developers.openai.com/api/docs/guides/your-data
ByteDance (Seedream) - see section 2.1 "Specific Terms for the BytePlus Model Services": https://docs.byteplus.com/en/docs/legal/docs-service-specific-terms#s-specific-terms-for-the-byteplus-model-services
Black Forest Labs may use submitted images for training purposes:
Black Forest Labs (Flux): https://bfl.ai/legal/flux-api-service-terms
Picture Instruments only forwards the images and does not store them at all.
All assets and generated images are immediately stored locally on your computer. However, for the actual image generation, the images must be sent to the respective AI providers.
That depends on the selected model. 1K models generate approximately 1 million pixels, 2K models around 4 million pixels and 4K models approximately 16 million pixels. The exact resolution also depends on the selected aspect ratio.
Yes. In this case, it is best to work with a selection. As long as the edited area is not larger than 4K, there is usually no loss in quality.
Yes. You can find your prompts including embedded reference images via the folder icon in your prompt library and easily share them.
No. The actual image generation runs on the servers of the AI providers. AI Lab handles the local organization, preview and workflow.
No. An internet connection is required for AI generation because the images need to be sent to external AI services. However, you can still use AI Lab offline for viewing images, preparing prompts and making brightness, saturation and contrast adjustments.
Yes. You can use existing images as references, edit the entire image or selectively edit individual parts of an image.
Yes. Using the selection feature, you can edit only the selected area. The rest of the image remains untouched.
Yes. You can generate multiple variations of an image simultaneously and compare them directly. You can also vary the prompts, the AI model and the aspect ratio if you want.
AI image generators are not fully deterministic. Therefore, even identical prompts can produce different results.
The speed depends on the selected model, the image resolution and the current workload of the respective AI provider.
Yes. You can use any images as style, subject or composition references.
Yes. Commercial use of AI Lab is possible. In addition, the terms of use of the respective AI providers apply, which you can find here:
Google (Nano Banana): https://ai.google.dev/gemini-api/docs/zdr?hl=en
OpenAI (GPT Image): https://developers.openai.com/api/docs/guides/your-data
ByteDance (Seedream): https://docs.byteplus.com/en/docs/legal/docs-service-specific-terms#s-specific-terms-for-the-byteplus-model-services
Black Forest Labs (Flux): https://bfl.ai/legal/flux-api-service-terms
AI Lab always saves the generated result image as jpg. However, the following formats can be opened:
png, jpg, jpeg, jif, jpe, gif, bmp, tif, tiff, psd, tga, targa, jp2, j2k, j2c, ico, hdr, cut, dds, exr, g3, iff, lbm, jng, jxr, wdp, hdp, koa, mng, pbm, pcd, pcx, pfm, pgm, pct, pict, pic, ppm, ras, sgi, wap, wbmp, wbm, webp, xbm, xpm
Yes. Prompts, reference images and generated results are stored locally on your computer and remain available even after restarting the software or cancelling your subscription.
Yes. AI Lab Standalone has an interface for Photoshop, allowing you to conveniently send images back and forth. In addition, there is the AI Lab UXP Photoshop plugin, which integrates natively into Photoshop after installation.
All images and prompts remain stored locally on your hard drive. You can access them at any time, even after your subscription has ended.
Gartenstrasse 3
51379 Leverkusen
Tel: +49 (0) 1579 25 10 500
