Edit Several Parts of One Image with FaithGPT's New Selection and Mask Tools
TL;DR: Mark several parts of an image with boxes, circles, or a brush. Give each numbered selection its own direction, then generate up to four variations while FaithGPT preserves the rest of the image.
Image editing gets frustrating when the tool understands the change but loses the location. You ask it to replace a vase and it changes the table. You ask it to remove a line of text and it redraws the whole wall. One prompt has to carry too much responsibility.
FaithGPT Image Studio now gives location and direction separate jobs. Mark the area first, then write what should happen there. You can do this several times on the same image before spending credits or starting the edit.
Several selections, several directions
Open an image you own, choose Edit , then open Edit selected areas . The editor supports up to eight selections in one pass:
Use a box around objects or sections with clear boundaries.
Use a circle around compact subjects such as a face, lamp, moon, or badge.
Use the mask brush for lettering, clothing, foliage, hair, or any shape that does not fit a neat outline.
Every selection gets a number and color. The direction beside number 1 belongs to number 1 on the image. Number 2 can ask for something different. You can resize or move boxes and circles, change the mask brush size, undo the last stroke, and clear a mask without disturbing the other selections.
Here is a practical edit you can send as one job:
Box the vase and write, "Replace the vase with a flowering olive branch."
Circle the ceiling light and write, "Change this to a warm paper lantern."
Paint over the old event date and write, "Remove this lettering and rebuild the wall texture."
FaithGPT keeps the clean source separate from a black-and-white inpainting mask. White pixels mark the complete area that may change; black pixels protect the rest of the image. The numbered directions are matched to their spatial bounds, while the colored controls you see in the editor stay in the interface and never become part of the visual source.
The edit runs as one composition. That matters. A lantern and an olive branch added in separate generations can drift in color or perspective. Processing the directions together gives the editor a better chance to match the same light, focus, grain, and scene geometry across the image.
Generate a few options at once
Image edits are nondeterministic: the same direction can produce a stronger subject, a cleaner blend, or a better texture on different runs. The editor therefore selects four results by default. You can choose one, two, three, or four before generating, and the button shows the total credit estimate for that batch.
All of the variations land together in the image's version strip. Compare them at full size, keep the strongest result, or continue editing from any version. If you only need a quick correction, choose fewer results and spend fewer credits.
Paint the part that needs work
Some changes need a mask. Text may curve across a poster. A power cable may cross a floor at an angle. A person may stand against a detailed background.
Choose Mask , set the brush width, and paint over the pixels you want changed. Add a direction such as "remove the cable and continue the wooden floorboards" or "change the blue jacket to dark green wool." FaithGPT converts those strokes into the white portion of the inpainting mask, then blends the replacement naturally at its edge.
A tight mask usually produces the cleanest result. Include enough of the old object for the model to understand what it must remove, but avoid painting across nearby faces, hands, or text that should stay untouched.
The rest of the Image Studio editor
Directed edit joins a workspace that already handles the parts of image production that tend to send people into separate apps.
Add and restyle text
Place text directly on the canvas, then drag, resize, rotate, align, or duplicate it. Plain text is free. Font faces and AI art styles show their credit cost before you apply them. Art styles can integrate lettering as neon, foil, chrome, or another material that follows the scene's light and perspective.
If a generated poster already contains lettering, Extract text layers removes the visible words from the base image and opens the detected phrases as movable layers. You can fix a date, rewrite a title, change the color, or move a line without regenerating the whole design.
Split the image into layers
Extract layers separates a flat image into movable image layers. Reorder them, hide one, resize a subject, rotate an object, or duplicate a layer. You can also add plain text inside the layer editor.
FaithGPT saves extracted layers with the image. Reopening the saved layer set is free because the expensive extraction has already happened. You only spend credits again if you ask for a fresh extraction.
Remove the background
Remove background cuts the main subject onto a transparent PNG. Use it for sermon slides, thumbnails, presentations, or a composite in Project Studio. The transparent result becomes another version in the same image set.
Upscale for larger output
Upscale creates a sharper, higher resolution version. The editor lists the available sizes and credit cost before it runs, which helps when you need a print file or a larger screen graphic.
Adjust light and color without spending credits
The Adjust panel changes brightness, contrast, saturation, warmth, and the tone curve. These controls are free and non-destructive. FaithGPT stores the values with the image, applies them in previews, and bakes them into a download. Reset the controls to return to the source.
Layer projects can apply those adjustments to one layer or the whole composition.
Move a still image into video
Choose Animate to carry an edited image into Video Studio. The image can become the starting frame for a short clip without another download and upload cycle.
Your original stays put
Applied edits create new versions inside the same image set. The source remains available in the version strip, so you can compare the result, continue from an earlier version, or try another direction.
The editors also share FaithGPT's portable .faithgpt.json project format. Export a project with its text layers, image layers, directed selections, and painted mask strokes. Import it later to continue the layout instead of rebuilding the edit from memory.
You can start with a FaithGPT generation or upload your own photo through Edit image on the Image Studio home page. Uploading the source is free. Any AI action shows its current credit cost before you confirm it.
The new selection tools are available in FaithGPT Image Studio . Open an image, switch to Edit , and choose Edit selected areas .
Helpful Scripture resources
Contact & support
Email the FaithGPT team at hello@faithgpt.io or visit the contact page .
Explore FaithGPT
