Start with a photo, product shot, sketch, or artwork. Tell ChatArt what to keep and what to change, choose an AI model, and generate the result you need.
Reference Image
Upload a reference image and describe what should look different. ChatArt uses the original image as visual context while generating the edited result.
Brighten a dark photo, shift the color palette, or give the scene warmer, softer, or more dramatic lighting.
Update clothing or accessories, adjust how a product looks, or add, remove, and replace visible objects.
Turn a photo or sketch into an editorial, cinematic, illustrated, anime, or another clearly described style.
Place a portrait, product, or character in a studio, outdoor scene, or another setting described in your prompt.
Use GPT Image 2, Nano Banana, Seedream, Flux, and other models in the same workflow. Try the same edit with another model and compare the results before choosing one.
Choose a creative direction, upload the image it requests, and make the result your own.
Upload a reference image, describe what to keep and change, then choose a model and generate.
Add a JPG or PNG reference image up to 30MB.
Explain what to keep, what to change, and the style or setting you want.
Select an image model and the available output settings, then generate your image.
Keep what already works, and change only what the new project needs.
Update a portrait or branded image so it fits a new post, cover, or recurring series.
Experiences with updating portraits, blog covers, and other everyday images.
Katie
Social media manager
I liked the portrait, just not the background for that post. I described a better one in ChatArt, and the result was much closer to what I needed.
Denys Kapush
SEO expertI use it for blog covers when the concept works but the look does not. I can try a different style or setting before settling on the final cover.
Meelad Mashaw
Social media managerI asked for warmer light and a cleaner background, then tweaked the prompt once. That was much easier than learning a full photo editor.
Gordon H.
Social media managerIf the first result feels off, I switch models. Some stay closer to the original while others push the style further, so I can compare before choosing.
Find practical guides for writing prompts, changing backgrounds, editing photos, and applying new styles.
Choose the right workflow, prepare your image, and know what to try when a result misses the mark.
Use image-to-image when you already have a subject, composition, or detail you want the model to reference. Use text-to-image when you want to create from a written idea without starting from an existing visual.
Choose an image where the subject and the details you care about are easy to see. A blurry, very small, or heavily cropped image gives the model less visual information to work with.
Name the exact features that should stay recognizable, such as a person's appearance, a product shape, or a character's clothing. Then describe only the parts you want to change. If the result still moves too far, make the instruction more specific or try another model.
You can request a targeted edit, such as changing the background, clothing, lighting, or one object. Because this is generative editing, nearby details may also change. Review the result and refine the prompt or generate again when needed.
Ask for fewer changes at one time, repeat the details that should stay recognizable, and remove instructions that compete with each other. You can also try the same image and prompt with another available model.
There is no single model for every edit. Try the same reference image and prompt with the available models, then compare which result best matches what you need. Available output settings are shown after you select a model.
The current Image to Image workflow supports JPG and PNG reference images up to 30MB.
Yes. After you register for ChatArt, your account receives free diamonds that you can use to try image generation. The diamonds required for a generation and any additional plan options are shown inside ChatArt.
Describe what to keep and what to change, then generate with ChatArt.
