AI Photo Editing Prompts That Produce Better Results

Author: AvocAIdo · Published: September 7, 2026

AI photo editing prompts are written instructions that tell an AI editor what to change, how the result should look and which details must remain untouched. Effective prompts combine a clear action, a precise editing target, visual direction and preservation constraints. This guide provides a reusable formula, copyable prompt library and practical methods for improving portraits, products and brand images. If you are new to the process, start with this step-by-step guide to editing photos with AI.

AI photo editing in short

• A reliable prompt covers 4 elements: the action, editing target, desired look and details that must remain unchanged.
• Focused prompts of 2 to 4 sentences are generally easier for an AI editor to interpret than long keyword lists.
• Portrait prompts should explicitly preserve identity, facial proportions, expression, skin texture, clothing and composition.
• A mask can limit an edit to 1 selected region while protecting the rest of the image.
• Stable Diffusion img2img users can treat denoising strength from 0.3 to 0.6 as a preservation-to-transformation control.
• Batch prompts should repeat the same identity, wardrobe, product geometry, palette and lighting constraints for every image.
• Content Credentials can document an image's provenance and editing history, but they do not prove that its subject matter is truthful.

To understand how AI photo editing actually works at a deeper level, see our companion explainer.

Checklist for writing an effective editing prompt

• Start with an editing verb such as remove, replace, extend, restore or recolor.
• Identify the exact person, object, background or masked region to edit.
• Describe the intended visual result in concrete photographic language.
• State which facial, product and compositional details must not change.
• Name the required lighting direction, softness and color temperature.
• Specify the intended aspect ratio or final publishing format.
• Use a mask or annotation when the edit concerns only one local area.
• Make one substantial change at a time, then refine with a follow-up prompt.
• Inspect faces, hands, text, logos, reflections and object edges before export.
• Remove sensitive metadata and visible identifying information before sharing files.

Methodology: This guide is based on current platform documentation, official provenance resources and 2026 prompt-writing references, with prompt examples rewritten into a consistent, tool-portable format.

How do AI photo editing prompts work?

An AI photo editor combines an existing image with a natural-language instruction. Unlike a text-to-image prompt, an editing prompt must describe both the desired change and the relationship between the new result and the source photograph.

A weak prompt says:

"Make this portrait look professional."

A stronger prompt says:

"Convert this portrait into a clean professional studio headshot. Use soft frontal lighting, a neutral light-gray background and natural skin texture. Preserve the person's identity, facial proportions, expression, hairstyle, clothing and original crop."

The second version gives the editor a target while limiting unnecessary changes. This is important because an editing model may otherwise reinterpret the entire frame, including parts the user already likes. Understanding multimodal AI models used by creative professionals helps explain why prompt specificity matters so much.

Picsart recommends a 4-part structure: action, what to edit, desired look and what to keep the same. It also describes 2 to 4 sentences as an effective prompt length. This is long enough to establish intent and constraints without burying the primary instruction. (zdroj: Picsart)

AI editing is normally iterative. Meta's guide to AI photo editing prompts advises users to adjust follow-up instructions and state what must remain untouched, such as facial expressions, clothing or the background. Instead of rewriting a full prompt after every result, identify the largest remaining error and correct that issue next.

For example:

"Keep the current face, expression, outfit, crop and background unchanged. Reduce only the orange color cast on the skin and restore a neutral daylight white balance."

This follow-up is more controllable than requesting another complete transformation.

What is the best formula for AI photo editing prompts?

A practical AI prompt for photo editing follows this formula:

Action + target + desired look + lighting and composition + preservation constraints + output requirements

Each component answers a different question.

Action: What should the editor do?

Begin with an unambiguous verb:

• Remove
• Replace
• Restore
• Recolor
• Extend
• Relight
• Retouch
• Reframe
• Sharpen
• Match

"Change the image" is vague. "Replace the background" defines an operation.

Target: Which part should change?

Name the target using its position, visual properties or relationship to another object:

• The woman in the foreground
• The red handbag on the table
• The sky above the mountains
• The wall behind the product
• The object inside the masked region
• The empty space on the right side

If several similar objects appear, include enough information to distinguish the intended one.

Desired look: What should the result resemble?

Describe observable qualities rather than subjective praise. "Make it amazing" gives the model no useful visual direction. Better terms include:

• Photorealistic studio portrait
• Clean editorial fashion photography
• Warm vintage film color
• Natural window light
• Matte product photograph
• Cinematic night scene
• Soft lifestyle photography
• Neutral e-commerce background

Use "ultra realistic" cautiously. Concrete information about texture, shadows, lens character and lighting usually provides more control than repeated quality adjectives.

Lighting and composition: How should the image be presented?

Specify the light source, direction and intensity:

"Use broad, soft window light from the left, gentle facial shadows and neutral white balance."

For composition, define the placement and crop:

"Keep the product centered with equal negative space on both sides and preserve the original camera angle."

Camera language can help communicate an aesthetic, but it should support the main edit rather than dominate it. An 85mm portrait look, for example, implies compressed perspective and limited background distraction. Do not add technical camera terms that conflict with the source photograph.

Preservation constraints: What must remain untouched?

This is the most frequently omitted part of a prompt. Both Meta and PromptSpace emphasize protecting facial features, composition and other original details. (zdroj: PromptSpace)

A reusable constraint block is:

"Preserve the original identity, facial proportions, expression, pose, wardrobe, product geometry, logo placement, camera angle and composition. Do not add new objects or text."

Choose only the constraints relevant to the image. A product edit may need strict geometry and label protection, while a portrait needs identity and skin-texture preservation.

Output requirements: Where will the image be used?

Finish with the intended format:

"Prepare the result for a square 1:1 social post, keeping important details away from the crop edges."

AvocAIdo displays 1:1, 16:9, 9:16, 3:4 and 3:2 aspect-ratio options in its generation workflow. Those formats cover common square, landscape, vertical and photographic compositions, but the final choice should follow the publishing channel.

How can you preserve faces, identity and references?

Identity drift occurs when the editor changes features that were not part of the request. Common symptoms include altered eye shape, facial proportions, age, hairstyle, expression or skin tone.

The solution is not simply to repeat "same person." Define the identity markers that matter, then separate them from the requested edit.

Identity-safe portrait template

"Edit only the lighting and background of this portrait. Preserve the same person's facial structure, eye shape, eye color, nose, mouth, jawline, age, skin tone, hairstyle and expression. Keep realistic skin texture and the original pose. Replace the background with a softly lit neutral studio wall."

Wardrobe-safe fashion template

"Apply a polished editorial fashion color grade to this full-body photograph. Preserve the model's identity, pose, body proportions, garment design, fabric texture, seams, accessories and exact wardrobe colors. Do not redesign or replace any clothing."

Reference-matching template

"Match the source image to the lighting, color palette and mood of the supplied reference. Keep the source subject, face, clothing, pose and composition unchanged. Transfer only the reference's soft side lighting, muted warm palette and subtle film contrast."

Reference images work best when their role is explicit. State whether a reference controls the identity, product, wardrobe, composition or visual style. Do not ask the model to "match the reference" without defining which properties should transfer.

For a campaign batch, use a fixed constraint block across every prompt:

"Maintain the same person, facial identity, hairstyle, cream jacket, gold earrings and natural makeup across the complete image series. Use the same warm beige palette, soft left-side key light and low-contrast editorial grade. Change only the setting and activity described below."

This repeated language acts as a production brief. It cannot guarantee identical results, but it reduces ambiguity and makes visual differences easier to identify during review.

AvocAIdo allows users to upload reference images and analyzes visual properties including color palette, lighting, mood and style. Its Brand Style Intelligence is designed to apply a reference-led visual direction across different subjects, although fidelity can vary with the model and complexity of the image.

Which portrait and headshot prompts can you copy?

These AI prompts for photo editing can be adapted by replacing the bracketed details. Remove brackets before submitting the prompt.

Natural portrait retouch

"Retouch this portrait subtly. Reduce temporary blemishes and uneven shine while preserving pores, fine lines, freckles, natural skin texture and the person's identity. Keep the original face shape, expression, hair, clothing, background and lighting."

Professional headshot

"Convert this image into a photorealistic professional headshot. Use a clean [light gray] studio background, soft frontal light and a polished but natural color grade. Preserve identity, age, facial proportions, hairstyle, expression and clothing."

Selfie correction

"Correct the close-range selfie perspective while keeping the person recognizable. Preserve the exact eyes, nose, mouth, jawline, skin tone and expression. Use natural daylight color, realistic skin detail and a balanced head-and-shoulders crop."

Background blur

"Create a subtle portrait-depth effect by softly blurring only the background. Keep the full face, hair edges, clothing and foreground sharply defined. Preserve the original lighting, colors, identity and composition."

Low-light portrait

"Improve this low-light portrait by reducing color noise, recovering natural facial detail and balancing the exposure. Keep the atmosphere realistic and retain genuine shadow depth. Do not brighten the scene into daylight or smooth away skin texture."

Corporate team consistency

"Match this team portrait to the approved headshot series. Use the same neutral background, crop, subject scale, soft lighting direction and natural color treatment. Preserve each person's identity, expression, hairstyle and clothing."

Editorial portrait

"Give this portrait a refined editorial finish with controlled contrast, soft directional light and restrained colors. Preserve the woman's identity, face, pose, wardrobe and the full original composition. Avoid synthetic skin, exaggerated makeup and added accessories."

For portraits, inspect eyes, teeth, hair boundaries, ears and hands at full size. If one area is wrong, request a local correction rather than regenerating the entire image.

What prompts work for fashion, lifestyle and travel photos?

Fashion and lifestyle edits need a balance between atmosphere and continuity. The prompt should create a clear visual world without changing the featured person, garment or location unnecessarily.

Clean studio fashion

"Transform the setting into a minimal fashion studio with a warm off-white backdrop and soft floor shadow. Preserve the model's identity, full pose, body proportions, outfit construction, exact garment colors and accessories. Keep realistic fabric texture and clean edges."

Magazine-style color grade

"Apply a sophisticated editorial color grade with controlled highlights, deep neutral shadows and subtle warm skin tones. Keep the face, makeup, wardrobe, background objects and crop unchanged. Do not alter garment colors or add text."

Outdoor lifestyle campaign

"Relight this lifestyle image with soft late-afternoon sunlight and gentle natural shadows. Preserve the same people, expressions, poses, clothes, location and camera perspective. Keep the result photorealistic and suitable for a cohesive brand campaign."

Food photograph

"Improve the food photograph with natural side lighting, accurate ingredient colors and appetizing but realistic texture. Preserve the plate, recipe, portion size, tableware and composition. Remove only distracting crumbs outside the plate."

Travel photograph

"Balance the exposure between the foreground and bright sky in this travel image. Recover realistic cloud and landscape detail without replacing the location. Preserve all people, buildings, signs, terrain and the original point of view."

Nature photograph

"Enhance natural detail and tonal separation in the landscape while keeping the scene believable. Preserve the original vegetation, mountain shapes, weather and composition. Avoid invented wildlife, oversaturated greens and artificial sky replacement."

Vintage lifestyle effect

"Apply a subtle vintage film treatment with restrained grain, gently softened contrast and slightly faded warm colors. Preserve the subjects' identity, wardrobe, setting and full composition. Do not add artificial scratches or alter facial details."

For lifestyle batches, repeat the same palette, contrast and lighting phrase in every prompt. A shared sentence such as "use soft natural light, muted earth colors and gentle film contrast" gives the image library a recognizable visual foundation. Learn more about how to make your AI-edited photos look professional and on-brand.

Which prompts work for products, mockups and brand visuals?

Product editing requires stricter constraints than mood-led creative work. Small changes to packaging dimensions, label text, logo placement or material finish can make an otherwise attractive image unusable. Having AI-ready brand style guide templates in place before writing prompts helps ensure every output stays within approved brand parameters.

Clean product background

"Replace the background with a seamless warm-white studio surface. Preserve the exact product shape, proportions, packaging, label, logo, typography, material and colors. Add only a soft realistic contact shadow directly beneath the product."

Product color correction

"Correct the photograph so the product matches the supplied color reference. Change only the color cast caused by the current lighting. Preserve the product geometry, surface texture, label, logo, printed text, highlights and shadows."

Lifestyle product placement

"Place the unchanged product on a clean stone bathroom counter in soft morning window light. Preserve its exact dimensions, packaging design, cap, label, logo and brand colors. Keep the product fully readable and use realistic contact shadows and reflections."

Packaging cleanup

"Remove dust, small scratches and distracting glare from the package. Preserve the original material texture, edges, folds, label design, typography, barcode and logo. Do not redraw, reinterpret or replace printed content."

Mockup replacement

"Place the supplied artwork inside the existing poster frame. Fit it accurately to the frame's perspective and preserve the artwork's layout, text, colors and proportions. Keep the room, frame, reflections, shadows and lighting unchanged."

Multi-image product consistency

"Match this product image to the approved campaign set. Use the same [warm gray] background, centered placement, product scale, shadow softness, exposure and color balance. Preserve the exact package geometry, brand colors, logo and label."

AI tools can distort text even when a prompt says to protect it. Compare every generated label against the approved artwork before publishing. For text-critical packaging, a practical workflow is to edit the photograph first and place the original label artwork during final compositing. See our guide to achieving visual consistency at scale with AI photo editing for more on managing large product libraries.

Understanding the must-have features in AI photo editing tools can help you choose a platform that supports the prompt techniques described here.

Print-ready prompt guidance

An AI prompt can request an appropriate composition, but it does not replace a print production file. Use the printer's supplied dimensions, bleed, safe area and color requirements rather than assuming a universal specification.

A useful editing prompt is:

"Extend the background to the supplied print canvas and continue its existing texture and lighting naturally through the bleed area. Keep the product, logo, headline and essential details inside the supplied safe area. Preserve generous edge space for trimming and use restrained colors suitable for later CMYK conversion."

After generation, create the final resolution, bleed and color profile in professional layout software. "High resolution" in a prompt is a visual request, not proof that the exported file meets a printer's technical requirements.

How do you restore or transform an old photo?

Restoration prompts should distinguish repair from invention. The goal is normally to recover visible information while retaining age, texture and historical characteristics.

Scratch and dust repair

"Restore this vintage photograph by removing dust spots, small scratches, fold marks and scanning artifacts. Preserve every person's identity, expression, clothing, hairstyle, pose and the original background. Retain natural photographic grain and period-appropriate contrast."

Faded photograph

"Recover balanced contrast and visible detail from this faded photograph. Keep the original monochrome character, faces, clothing, objects and composition. Avoid modern sharpening, synthetic skin and invented facial details."

Conservative colorization

"Colorize this black-and-white photograph conservatively. Use natural skin tones and historically plausible, muted colors while preserving the original faces, clothing patterns, lighting, grain and composition. Do not add or remove objects."

Damaged facial area

"Repair only the damaged section of the face using the surrounding visible features as reference. Preserve identity, age, expression, head angle, lighting and photographic texture. Do not beautify or modernize the person."

Creative transformations require different language because reinterpretation is intentional.

Cinematic transformation

"Transform the lighting and color grade into a cinematic night scene with cool ambient light and a warm practical light source. Preserve the person's identity, pose, wardrobe, location geometry and camera angle. Keep the result photorealistic."

Fantasy portrait

"Reimagine the environment as an elegant fantasy forest while preserving the same person, face, expression, pose and clothing silhouette. Integrate soft atmospheric light and realistic depth. Do not add extra people or alter facial anatomy."

Anime interpretation

"Translate the portrait into a refined anime illustration while retaining the subject's recognizable hairstyle, expression, pose, outfit colors and composition. Use clean linework, controlled shading and consistent facial proportions."

Science-fiction scene

"Transform the existing city background into a plausible science-fiction setting with restrained architectural details and atmospheric lighting. Preserve the foreground subject, identity, wardrobe, pose and perspective. Avoid added text, logos and unnecessary characters."

How do masks improve local AI edits?

A mask identifies the image region available for regeneration. It is often more reliable than asking a prompt-only editor to infer which object should change.

AWS documentation explains that a maskPrompt can describe elements to change during inpainting or elements that must remain untouched during outpainting. Masking is particularly useful for removing a small object, repairing one facial area or changing a garment detail without rebuilding the full frame. (zdroj: AWS)

Use this general mask-aware template:

"Inside the masked region only, [describe the edit]. Match the surrounding perspective, light direction, color, texture, grain and depth of field. Keep every unmasked region unchanged. Do not modify the face, clothing, background or composition outside the mask."

Examples include:

"Inside the masked region only, remove the plastic bottle and reconstruct the grass behind it. Match the surrounding grass texture, sunlight and focus. Keep all unmasked pixels and objects unchanged."

"Inside the masked region only, change the shirt from blue to dark green. Preserve the fabric folds, stitching, shadows and highlights. Do not alter the face, hands, jacket, background or crop."

"Inside the masked region only, repair the torn corner of the vintage photograph. Continue the existing wall texture and photographic grain. Do not sharpen or recolor the rest of the image."

When expanding a canvas, distinguish protected content from the area being generated:

"Extend only the new canvas area on the left and right. Continue the existing background, horizon, lighting and depth naturally. Keep the original image region, person, face, product and composition unchanged."

Local edits still require inspection along mask boundaries. If the new material does not blend correctly, use a follow-up prompt that addresses edge texture, perspective or light direction rather than repeating the original request.

How should prompts be adapted for different AI editors?

The underlying prompt structure is portable: identify the action, target, visual result and protected details. What changes between platforms is the available control method. For a comparison of the best AI photo editing apps ranked by features and price, see our dedicated review.

Conversational editors

Chat-based editors support follow-up instructions. Begin with a focused 2 to 4 sentence request, inspect the result and correct one issue at a time.

A useful follow-up is:

"Keep the current version, but restore the original eye shape and natural skin texture. Do not change the background, crop, clothing or lighting."

Mask-based editors

Place the editing instruction first and refer directly to the mask:

"Replace only the masked wall area with pale textured plaster. Match the room's warm side light and perspective. Preserve all unmasked regions."

Some tools use painted masks, while others accept a natural-language mask description. Follow the editor's interface rather than adding unsupported syntax.

Reference-based editors

Explain the role of each uploaded image:

"Use image 1 for the person's identity and clothing. Use image 2 only for lighting, palette and mood. Preserve the composition of image 1 and do not transfer people or objects from image 2."

This prevents the editor from treating all references as interchangeable content sources.

Stable Diffusion img2img

PromptSpace recommends a denoising strength from 0.3 to 0.6 as a balance between preservation and creativity. Lower values retain more source structure, while higher values allow a stronger transformation.

Change this setting gradually and keep the prompt constant while comparing results. Otherwise, it becomes difficult to determine whether the wording or the parameter caused the difference.

Multi-model workflows

Different image models can interpret the same prompt differently. Preserve the core brief across all runs, including the subject, requested change, style and constraints. Compare identity retention, product accuracy, lighting and composition rather than selecting a result only because it looks dramatic.

AvocAIdo runs Grok Imagine, Seedream 5 Lite, NanoBanana 2 and GPT Image 2 in parallel for each generation, allowing users to review 4 interpretations. Its ArtDirector AI converts plain-language descriptions into model-optimized prompts and is listed as a free prompt-enhancement feature. (zdroj: AvocAIdo)

Why do editing prompts produce poor results?

Most failures can be traced to conflicting instructions, an unclear target or missing preservation constraints.

The whole image changes

Cause: The prompt describes the desired atmosphere but does not limit the scope.

Correction:

"Apply the requested change only to the background. Preserve the person, face, hair, clothing, pose, foreground and crop unchanged."

Use a mask if the editor supports one.

The face no longer matches

Cause: A broad transformation has overridden identity details.

Correction:

"Restore the source identity and exact facial proportions. Keep the current background and lighting, but match the original eyes, nose, mouth, jawline, age, skin tone, hairstyle and expression."

Products or logos become distorted

Cause: The model is regenerating the package instead of editing its environment.

Correction:

"Treat the product and all printed packaging as locked source content. Change only the background and contact shadow. Do not redraw the label, logo, typography, barcode or product edges."

The result looks artificial

Cause: The prompt requests excessive smoothing, sharpening, saturation or vague "perfect" quality.

Correction:

"Reduce the retouching intensity. Restore realistic pores, fine hair, fabric texture, natural shadow variation and restrained color. Avoid plastic skin, glowing edges and excessive sharpness."

A batch looks inconsistent

Cause: Each prompt uses different language for palette, light and composition.

Correction: Create a fixed batch specification containing the same identity, wardrobe, product, palette, lighting direction, contrast and crop. Change only the scene-specific sentence.

The requested object is not removed cleanly

Cause: The editor lacks information about what should replace it.

Correction:

"Remove the sign inside the masked area and reconstruct the brick wall behind it. Continue the existing brick pattern, mortar lines, perspective, shadows and focus."

Too many errors appear at once

Cause: The prompt combines several structural transformations.

Correction: Divide the task into stages. Handle background replacement first, then lighting, color and small retouching through separate edits. This makes each result easier to evaluate and reverse.

How can edited photos remain private and transparent?

Before uploading a photograph, check both embedded metadata and visible information. EXIF metadata may contain GPS, timestamp and device details, while the image itself may expose license plates, badges, documents, house numbers or identifying reflections.

A privacy-focused prompt can help with visible content:

"Remove the license plate inside the masked area and replace it with a plain, non-identifying plate surface. Match the vehicle's perspective, reflections and lighting. Preserve the rest of the vehicle and all unmasked regions."

An AI edit is not a substitute for a complete privacy review. Examine reflective windows, mirrors, computer screens, name tags and the full edge of the frame at 100 percent size. The Imgezy privacy guide also recommends stripping EXIF metadata before sharing. (zdroj: Imgezy)

For provenance, C2PA Content Credentials can record where a file came from, which compatible software handled it and how it was edited. Google reports using C2PA Content Credentials across a growing number of its generative media tools. (zdroj: Google)

The exact verification steps published by the Canadian Association for Photographic Art are:

1. "Visit contentcredentials.org/verify"
2. "Upload or drag-and-drop your image"
3. "Review the complete authenticity report"

(zdroj: CAPA Canada)

A valid credential can show traceable origin and editing history. It cannot establish whether a photograph was staged, captioned accurately or used in the correct context. Material documentary edits should therefore be disclosed clearly, especially when objects, people, environments or events have been added or removed.

AvocAIdo tip

Create a short visual brief before generating variations: define the protected subject, approved palette, lighting direction, composition and intended aspect ratio. AvocAIdo can analyze uploaded references for color palette, lighting, mood and style, then generate through 4 AI models in parallel. Use ArtDirector AI and Brand Style Intelligence to convert a plain-language idea into optimized prompts and retain reusable visual direction across projects. Review all outputs against the same identity, product and brand constraints before selecting the strongest version.

Frequently asked questions

What should I include in a photo editing instruction?

Include 4 core elements: the action, exact target, desired appearance and details that must remain unchanged. Keep the instruction focused, with 2 to 4 sentences covering lighting, composition or output format only when they directly affect the edit.

How do I stop AI from changing a person's face?

Name the identity features that must be preserved, including facial proportions, eyes, nose, mouth, jawline, age, skin tone, hairstyle and expression. If only 1 area needs correction, mask that region and state that the unmasked face and composition must remain unchanged.

Can I use the same prompt in ChatGPT and Gemini?

The same action, target, style and constraint structure can be used across conversational editors, although results will differ by model. Start with 2 to 4 focused sentences, then use one follow-up instruction at a time to correct identity, lighting or background errors.

What is a good prompt for professional product photos?

Request a clean background, realistic contact shadow and accurate color while locking product geometry, packaging, label, typography and logo. Review every piece of generated text against the original, because even a visually strong result may contain altered characters or spacing.

Should I use negative prompts?

Use concise constraints that address likely errors, such as "do not alter the face, logo, wardrobe or composition." A short list of 3 to 5 relevant restrictions is generally clearer than a long block of unrelated negative keywords.

How can I keep a campaign visually consistent?

Repeat one fixed specification across the whole batch, covering identity, wardrobe, product, color palette, lighting direction, contrast and crop. Change only the subject action or setting, then compare at least the selected outputs against the same approved reference image.

What denoising strength should I use for img2img?

A cited working range for Stable Diffusion img2img is 0.3 to 0.6. Begin near 0.3 when source preservation matters, then increase gradually toward 0.6 when the requested visual transformation is not strong enough.

Can an AI prompt make an image ready for print?

A prompt can request the correct composition, edge space and background expansion, but it cannot certify a print-ready file. Apply the printer's supplied dimensions, bleed, safe area, resolution and color profile in layout software after generating the edited image.

Next step

Choose one source image and write a 4-part brief covering the edit, target, desired look and protected details. Then start with 2,000 free credits on AvocAIdo, with no credit card required, and compare the 4 parallel model outputs against the same visual brief. You can also explore AvocAIdo's pricing plans to find the tier that fits your workflow.

Zdroje

1. Meta, AI Photo Editing Prompts
2. Picsart, AI Photo Editing Prompts
3. AWS, Prompting with Image Masks
4. Google, Identifying AI-Generated Media Online
5. CAPA Canada, New AI Authentication Standards
6. PromptSpace, AI Photo Editing Prompts 2026