# FLUX Kontext Komposer Guide: Presets and Playground Edits > Published 2026-04-04, updated 2026-09-05 · https://www.promptzone.com/ellis_abbott/flux-kontext-composer-ai-image-tool-launch-14p3 --- title: "FLUX Kontext Komposer Guide: Presets and Playground Edits" published: true description: "Understand FLUX Kontext Komposer presets, use the documented BFL Playground editing workflow, and review changes to faces, products, and text." tags: ai,imagegeneration,prompting,tutorial --- Black Forest Labs introduced FLUX Kontext Komposer as a hosted image-editing experience with presets for scene changes, relighting, product placement, and posters. For a currently documented editing workflow, use Kontext pro or max in the BFL Playground: upload an image and describe the change. The current help page does not confirm availability of the original Komposer presets. [1], [2] ## What are the key facts about Kontext Komposer? | Fact | Verified detail | | --- | --- | | Developer | Black Forest Labs. [1] | | Released | Komposer and Kontext-powered presets were announced July 11, 2025. [1] | | Type | Hosted preset-based image-editing interface. [1], [2] | | Size or parameters | Not published for Komposer in its announcement; it is not presented as a standalone checkpoint. [1] | | License and access | Playground service access; no open weights are published for Komposer, and hosted pro models have no open weights. [1], [3], [4] | | Where it runs | Hosted browser interface at launch; current help documents Kontext editing in BFL Playground. [1], [2] | ## Which edits did Kontext Komposer presets support? The Komposer launch grouped common transformations into presets: new settings or styles, relit headshots, product placements, and movie posters. These categories provide a starting point for choosing an edit. [1] For a product example, start with a reference image whose important features you can name. List the outline, color, material, and visible label before requesting a different setting. That short description becomes your review guide. A generated scene may fit the desired mood while failing to preserve a handle, a seam, or a detail on the packaging. Kontext's underlying capabilities include targeted edits, character consistency, and reference-style transformations. BFL's model announcement also describes building on an image through successive changes. [5] ## What are the limits of Kontext preset editing? The Komposer announcement describes its launch features. BFL's current Playground help instead documents model selection and prompt-based editing; it does not guarantee that every original preset name remains available. [1], [2] Use the controls actually provided by the service. If a named preset is absent, the documented route is to upload the image and describe the desired transformation in text. [2] Kontext's technical report includes failures in identity preservation, missed instructions, and artifacts after repeated edits. Those are relevant limits when evaluating an image produced through a preset interface. [6] Keep the original image available. If a later revision changes something essential, return to the source rather than assuming another broad transformation will restore the missing detail. The current Kontext support page describes roughly one-megapixel outputs. A visually successful composition still needs inspection at the size and crop required for its intended placement. [7] The Playground currently uses the same pricing as BFL's API, according to its help documentation. Check the displayed model and current credit requirements before planning repeated generations. [2] Komposer is not a local model download. BFL separately publishes Kontext dev weights, which require their own inference setup and non-commercial model license or appropriate commercial rights. [3], [8] ## How do you make Kontext image edits in the BFL Playground? Use the hosted BFL Playground for the documented Kontext editing workflow. [2] The underlying models have a separate API; these sources document no standalone Komposer CLI. [2], [4] ### Start from a clear editing goal Open the Playground and choose Kontext pro or max. Upload your source image and describe the intended edit, then generate and inspect the returned image. These are the steps in BFL's current help article. [2] For a background change, try: “Place this backpack on a wooden bench in a garden. Keep its shape, fabric color, straps, and front pocket unchanged.” This is an example brief, not an official preset command. Check the backpack against the source before judging the garden. If the product is wrong, narrow the edit or choose another result instead of approving the scene because its lighting is pleasing. For portrait lighting, describe the effect you want in ordinary language: “Use soft window light from the left. Preserve the face, hairstyle, pose, and framing.” Keep your original alongside the generated version. For a poster, decide on the exact wording before the image edit. Specify the title, its position, and which part of the source image should remain unobstructed. ### Review and save a selected result Check the requested transformation first. Then examine the properties that were meant to remain fixed, including recognizable facial features, object geometry, and any existing printed information. Write a short note for each rejected result. “Wrong zipper placement” or “changed eye shape” tells you what failed and provides a concrete target for the next attempt. BFL documents a download button at the top right of selected generated content. Save the accepted file and retain a separate copy of the source image. [9] Keep the model choice and edit instruction with your output. If a preset was available and used, record its displayed name as part of the experiment rather than relying on memory. ### Move a repeatable edit into an application The Playground help describes an `API Code` control on a successful generation. It provides code in the selected language using that generation's parameters, which you can use as an integration starting point. [2] Check that the exported code selects the model you intended. Treat the resulting application as a separate integration that needs its own request handling and service credentials. The [Kontext overview](/riya_ahmadi/flux-kontext-ai-model-debuts-3988) explains the hosted and downloadable variants. Use it when the next decision concerns model access rather than visual preset selection. ## How do Komposer presets compare with the Kontext API and dev? | Route | What you control | | --- | --- | | Komposer's launch presets | A transformation category selected in the hosted interface described at launch. [1] | | Kontext API | A named model endpoint, input image, instruction, and documented parameters. [4] | | Kontext dev | A local editing checkpoint with its own runtime and license. [8] | The [ComfyUI pillar](/tomas_novak/comfyui-2026-the-complete-guide-to-power-user-ai-image-generation-1g17) covers the workflow concepts involved in a local implementation. Keep that deployment choice separate from deciding whether an edit looks right. ## What should you know before using Kontext presets? ### Is Kontext Komposer a downloadable model? Kontext Komposer was announced as a hosted editing experience with presets. Kontext dev is a separate downloadable checkpoint with its own license. [1], [8] ### Do I have to write a prompt? Kontext Komposer presets were introduced for transformations without written prompts. BFL's current help documents text instructions as the supported Playground editing workflow. [1], [2] ### Can I still use a particular launch preset? BFL's current Playground help does not confirm availability of every original Kontext Komposer preset. Use the available interface or describe that transformation through the documented editing flow. [2] ### Will a product remain exactly unchanged? Kontext edits can alter details that were meant to stay fixed. The model's technical report documents identity loss and missed instructions, so compare the edited product with the source image. [6] ## Sources - [BFL's original Komposer announcement, retrieved through its official post embed][1] - [Current BFL Playground documentation][2] - [Official repository and open-weight distinction][3] - [Kontext API editing documentation][4] - [Kontext model launch][5] - [Kontext authors' technical report][6] - [Current Kontext capabilities and output resolution][7] - [Kontext dev release and license][8] - [Playground download instructions][9] [1]: https://publish.twitter.com/oembed?url=https%3A%2F%2Fx.com%2Fbfl_ai%2Fstatus%2F1943635700227739891 [2]: https://help.bfl.ai/articles/8667153955-what-is-the-bfl-playground [3]: https://github.com/black-forest-labs/flux [4]: https://docs.bfl.ai/kontext/kontext_image_editing [5]: https://bfl.ai/blog/flux-1-kontext [6]: https://arxiv.org/html/2506.15742v2 [7]: https://help.bfl.ai/articles/5186006235-what-is-flux-1-kontext [8]: https://bfl.ai/blog/flux-1-kontext-dev [9]: https://help.bfl.ai/articles/7012361885-how-do-i-download-generated-images-on-the-playground ## Related guides on PromptZone - [Realistic Photos with FLUX](/stabletom/realistic-photos-with-flux-57aa) - [Best SDXL Models in 2026](/tara_suzuki/best-sdxl-models-in-2026-realistic-anime-and-all-purpose-checkpoints-116) - [The Ultimate Guide to Fooocus Image Prompts](/jj_ai/the-ultimate-guide-to-fooocus-image-prompts-1759)