Visual Editing

Nano Banana 2.1 adds mask editing, stronger subject consistency, bug fixes, and output up to 4K at a lower price. The model is available through an API and appears in Higgsfield, Pika, and ElevenCreative.

Venturetwins reports making nine successive FLUX 3 edits to a Chloe ad without visible degradation. Magnific offers the model with multiple image references, element placement controls and targeted detail editing.

AIandDesign reports remaking Deadfall's SNES sprites through Scenario's MCP in Codex, which also generated artwork-review documents. The creator shared an upgraded ROM for download, with footage running in BSNES.

FLUX 3 Image adds bounding-box layouts, precise multi-turn edits, 4K output and support for up to 10 references. A creator comparison found it most accurate at following fine-detail editing instructions.

Magnific publishes six poster workflows made with Ideogram 4.5. Each starts with a base image without text, then uses edits to add collage, illustration or typography.

Higgsfield released Claude Opus 5.5 skills for arranging editable Blender scenes, sorting footage, cleaning shots, compositing scene changes, and directing TouchDesigner. The release also adds a Vectorize skill that converts raster illustrations into layered outputs.

Midjourney updated its editor with selected-pixel inpainting, a newer edit model replacing Omni-Reference, revised tile behavior, and an early live preview. The preview is still an early prompt test, while the newer model replaces Omni-Reference in edits.

Higgsfield’s demo shows GPT-6 Astra handling camera work, 3D rendering, motion blur, grain matching, and compositing for a live-action VFX shot. A separate demo claims Astra retains context in After Effects.

Higgsfield says GPT-6 Astra created 120 butterfly-animation frames with 600 editable Photoshop layers. The process used GPT Image 2.5 frames and handmade animation references over five to six hours.

Midjourney says its new edit model improves character sheets and character consistency. Company replies say the model replaces Omni and should outperform the prior model.

GPT Image 2.5 adds the drawing-led Sketch interface plus faster Flare and precision-focused Sunburst API variants. Early creator tests report stronger edit persistence, text rendering, lighting, and reference adherence, though the source summary ends mid-caveat.

Creators report Astra writes scripts and carries out edits in Adobe apps. Demonstrations show it operating Photoshop, Illustrator, After Effects, and Krita through desktop or MCP workflows.

Topaz for Web brings video enhancement to a browser app, including upscaling, slow motion, frame interpolation, and SDR-to-HDR conversion. It also includes image upscaling and deblurring.

A creator used Midjourney V8.2 Edit for prompt-directed splash retouching, two-image application shots, and liquid-product composites. One drip shot combined liquid images across three edit steps.

Midjourney updated its V8.2 Edit test model with instruction edits, up to four image references, 2x --hd output, inpainting, and outpainting. Midjourney says the update improves image quality and is seeking feedback.

Midjourney says V8.2 adds instruction-based edits, image inputs, brush inpainting, and outpainting. The tools retain personalization, moodboards, and style references.

Creators posted Grok Imagine Image 2.0 demos for object-segment edits, reference blending, text rendering, resizing, and background removal. The samples focus on posters, covers, product-style images, and diagrams.

Designers posted reusable prompt systems for jelly, frozen-glass, and collectible badge treatments that turn brand marks into product-ready assets. The evidence includes variable-driven prompts, image-editing tips, and examples for posters, icons, and campaign visuals.

Users say Google Earth's AI image tool disappeared after about 36 hours. Bilawal Sidhu said the Nano Banana test was rolled back after debate over generative images inside an evidentiary map layer.

Dreamina users report Seedance 2.5 is live globally with native 30-second clips, long-form extension, up to 50 multimodal inputs, local edits, green screen, and 3D white-model controls. Higgsfield and Luma previewed integrations.

Creators show Google Earth using Nano Banana to reskin homes, cities, and map views with styled image generation. Users also report no obvious download or export control in the current experience.

Luma says Layers is live in Luma Agents for extracting editable elements from uploaded or generated images. Demos show product swaps, headline edits, object inserts, and modify-video keyframes.

Shared resources included design-skills repos, a UI visual dictionary, CRM table tokens, and Claude Opus 5 website prompts. The reusable context can speed front-end generation but does not replace product-specific design choices.

Midjourney says V8.2 is now the default image model after a short test cycle, with updates to aesthetics, personalization, and image quality. David Holz singled out personalization as a strength.

Magnific says its Photoshop plugin adds 4K upscaling, relighting, camera control, skin enhancement, and searchable access to Magnific creations and stock assets. The controls run inside the Photoshop editor.

DecartAI and Marc Kohlbrugge showed Lucy 2.5 restyling live video into new characters, outfits, environments, and styles from text, video, or Nano Banana references. One listing quoted $0.04 per second for real-time filters.

Linus Ekenstam showed Seedream 5.0 Pro edits in BytePlus Lumina using reference photos and art-direction marks. The demos covered posters, character sheets, items, animals, blueprints, and Seedance references.

Leonardo, BytePlus Lumina, and Magnific posts showed Seedream 5.0 Pro handling multilingual text, dense infographics, marked-up edits, and surveillance-fashion prompts. BytePlus was cited as an official API path.

Magnific added Seedream 5.0 Pro with camera settings and later promoted unlimited use in select regions. Leonardo users also tested the model across 10 styles, including posters and character sheets.

Runway, Pika, Magnific, and Promptsref added Seedream 5.0 Pro access. Magnific advertised unlimited 1.5K generations for a month, while Runway highlighted reference images and text in up to 14 languages.

Creators report Gemini Omni Flash is now available inside Adobe Firefly Boards. Early tests focused on video restoration and restyling prompts, with broader mobile rollout mentioned in replies.

Meng To showed Taste with Codex or Claude Code rebuilding landing pages from screenshots and videos. Voila also launched a tool that inspects UI and turns edits into code, copy, prompts, tasks, or agent handoffs.

Kai Gani described using Sora 2 castings for character-cameo references, then reported Codex/GPT 5.5 caught an anatomy-placement error in generated video frames. The check used contact sheets and still frames, not audio or motion.

Magnific released a Photoshop plugin for generation, upscaling, retouching, background removal, relighting, reframing, feed browsing, and stock search. Try it for image-editing workflows; same-day demos showed Gemini Omni motion graphics and Seedance 2.0 character-sheet consistency inside Magnific.

Stages Pro added Replace All, a shot-recasting tool that swaps characters, cast, objects, or locations while preserving framing, lighting, shadows, grade, and composition. Try it for recasting shots, and watch related EDITX tutorials and character-consistency workflows.

Photoshop AI Assistant was presented as a text-driven editor that targets specific parts of an image, preserves consistency, and names layers automatically. Watch how quickly it speeds selective edits inside Photoshop and whether users trust it for more complex retouching.

NVIDIA released SANA-Streaming as a system for real-time editing of minute-long video, including scene swaps, style transfer, inpainting, and robotics overlays. The demo emphasized temporal consistency across longer clips rather than one-off image transformations.

Magnific rolled five new image, video, and upscaling models into its workspace and added Auto Layers for editable text, subject, and background separation. Designers can now move from layout to layered edits and model switching inside one tool, while MCP generations still use Magnific credits.

Magnific added a Designer node inside Spaces, letting users place generated images and text directly into layouts with fonts, effects, and multi-page canvases. The update also extends Auto Layers editing across Designer, Spaces, and MCP.

Magnific turned its MCP connector live on all paid plans and added a Slack-to-image workflow with auto layers and resize exports. The rollout matters because image generation, editing, and asset reuse can stay inside chat-driven creative workflows, with credits consumed per MCP request.

Rendergeist Pulse added live audio-reactive playback, TV casting, more effects, and mobile and iOS support after its free launch. Posts also said the tool stays client-side and can animate uploaded video or image files against music, so creators can try it for DJ visuals.

Reve published a technical explanation of Large Layout Models, arguing that image generation works better from structured layouts than prose-only prompts. Try layout-first editing for granular changes to hair, lipstick, and skin tone without rebuilding the whole image.

Runway opened Edit Studio, putting Aleph 2.0’s localized video edits in the main app so users can change only selected parts of a shot. Watch for render-to-video refinements and agency workflow savings as the tool rolls out.

Vitrine opened a TestFlight beta for turning photos into glass-style wallpapers with live Metal previews, presets, and high-resolution export. Use the plain-text recipes to reuse settings while keeping processing local on the Mac.

Runway put Aleph 2.0 on its API for up to 30 seconds of 1080p editing across multi-shot sequences while changing only selected elements. A same-day creator test inside Adobe Firefly suggests early demand, so watch for broader end-to-end editing workflows.

Sponsored creator threads showed Wonder 3 improving portrait detail, skin texture, lighting, and text clarity from rough inputs without stacking multiple enhancement passes. The examples covered photo portraits and low-resolution renders across desktop and web tools, so compare outputs before adopting it in production.

Topaz launched Wonder 3 in Image Web as a broader realism enhancer for both clean and degraded images, with low, medium, and high strength controls. Test it on stylized characters to see whether it sharpens texture and detail without flattening the look.

Adobe Firefly AI Assistant entered public beta with creator demos focused on automatic crop, framing, and sizing changes from one image across multiple platforms. That turns a previously announced resizing feature into a live assistant workflow inside Firefly.

A paid-partner walkthrough showed Topaz Photo's new Autopilot analyzing a portrait and recommending Denoise plus Noise-Aware Sharpening from one toggle. The feature shifts cleanup toward guided first passes, though today's direct evidence comes from sponsored creator demos.

Runway launched Aleph 2.0 inside a new Edit Studio that lets users change one frame, preview it, and spread the edit across a whole video. The release gives web editors finer control than clip-by-clip regeneration.