Contents<\/h2>\n\n- Test setup and documented run data<\/a><\/li>\n
- What the six outputs show<\/a><\/li>\n
- When to pick Flux Kontext Max Multi<\/a><\/li>\n<\/ul>\n
Model and test setup<\/h2>\n
The model page is Flux Kontext Max Multi on Wiro<\/a>. Each test passed two image URLs: the first was the base scene and the second was the reference. Prompt upsampling was set to false<\/strong>, safety tolerance to 2<\/strong>, seed to 42<\/strong>, and output format to JPEG<\/strong>. Keeping prompt upsampling off matters here: the results reflect the direct instructions shown in each caption rather than an expanded prompt.<\/p>\nWiro’s model documentation shows a completed example with 6.0 seconds<\/strong> elapsed time. It does not publish a fixed per-output price. A separate cancelled-task example lists a total cost of 0.003510<\/strong>, but that is not evidence of a delivered-output price, so it should not be used to budget a production run. Queue time, inputs, and platform settings can also change the elapsed time.<\/p>\nFor background on the FLUX family and its API concepts, see the Black Forest Labs documentation<\/a>. For adjacent editing workflows, compare this test with Qwen Image Edit Fast<\/a>, Seedream V5 Lite image-editing tests<\/a>, and GPT Image 1.5 text-edit tests<\/a>.<\/p>\nWhat the six outputs actually show<\/h2>\n1. Logo on a shirt: reference fidelity with a photographic base<\/h3>\n![\"Base](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-a1.jpg\")
Input 1: the base portrait.<\/figcaption><\/figure>\n![\"Logo](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-a2.png\")
Input 2: the logo reference.<\/figcaption><\/figure>\n![\"Flux](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-out-1.jpg\")
Prompt: Put the logo onto the man’s shirt with small size.<\/figcaption><\/figure>\nThe output places a large illustrated koala-and-bottle mark on the right side of the white shirt. The man, pose, shirt buttons, and shop-like background remain recognizable. The edit is visually clean, though the result is larger than the prompt’s \u201csmall size\u201d request. This is a useful reminder that placement and scale need separate, explicit constraints when brand artwork matters.<\/p>\n
2. Hair-look transfer: the edit changes more than hair<\/h3>\n![\"Base](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-b1.jpg\")
Input 1: the base portrait.<\/figcaption><\/figure>\n![\"Hair](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-b2.jpg\")
Input 2: the hair reference.<\/figcaption><\/figure>\n![\"Flux](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-out-2.jpg\")
Prompt: Make the woman’s hair look like it does in the second image.<\/figcaption><\/figure>\nThis output has short, swept brown hair, but it also changes the subject’s face, outfit, and framing. It succeeds as a loose visual transfer, not as a strict retouch. Pick this approach for mood boards or concept exploration. For a client headshot, state the protected areas directly: keep face, identity, clothes, pose, and background unchanged; change only hair color and style.<\/p>\n
3. Wall pattern transfer: geometry holds, reference texture does not<\/h3>\n![\"Room](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-c1.jpg\")
Input 1: the room.<\/figcaption><\/figure>\n![\"Pattern](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-c2.jpeg\")
Input 2: the pattern reference.<\/figcaption><\/figure>\n![\"Flux](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-out-3.jpg\")
Prompt: The first image is the room. The second image is the pattern. Paint the wall behind the bed with the pattern in the second image.<\/figcaption><\/figure>\nThe bed, windows, floor, and wall position survive the edit. The rear wall becomes a dark, aged brick surface rather than a close reproduction of the supplied pattern. That makes the result convincing as an interior redesign concept, but weak for a product-accurate wallpaper preview. Use it to explore surface direction; use a compositing workflow when a specific repeat pattern must remain exact.<\/p>\n
4. Logo on an illustrated shirt: strong text and local placement<\/h3>\n![\"Stylized](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-d1.jpg\")
Input 1: the stylized portrait.<\/figcaption><\/figure>\n![\"Logo](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-d2.png\")
Input 2: the logo reference.<\/figcaption><\/figure>\n![\"Flux](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-out-4.jpg\")
Prompt: Print the logo in small size at the center of the woman’s shirt.<\/figcaption><\/figure>\nThe output keeps the selfie-like illustration and puts a readable green \u201cWIRO\u201d wordmark across the shirt. It is centered and local, although again not especially small. Among these outputs, this is the clearest evidence that the model can carry a simple brand mark into a stylized scene without turning it into gibberish.<\/p>\n
5. Sleeve patch: localized edit with a changed subject<\/h3>\nInput 1: the reused base portrait.<\/figcaption><\/figure>\nInput 2: the reused logo reference.<\/figcaption><\/figure>\n![\"Flux](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-out-5.jpg\")
Prompt: Add the logo from the second image as a small patch on the left sleeve of the woman’s hoodie. Keep everything else the same.<\/figcaption><\/figure>\nThe model produces a black hoodie with a tidy koala patch near the left chest\/shoulder area. The patch itself looks integrated, but the person and clothing change from the starting portrait. This is a good direction test for merch mockups, not proof that the model will preserve a supplied talent photo. In production, run one change at a time and reject any result that alters the protected subject.<\/p>\n
6. Remove and move: two instructions complete with a visible compromise<\/h3>\n![\"Tea](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-f1.jpg\")
Input 1: the kitchen scene.<\/figcaption><\/figure>\n![\"Edge](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-f2.jpg\")
Input 2: the guidance reference.<\/figcaption><\/figure>\n![\"Flux](\"https:\/\/wiro.ai\/blog\/wp-content\/uploads\/2026\/03\/flux-out-6.jpg\")
Prompt: Remove the tea glass. Put the bird left of the teapot.<\/figcaption><\/figure>\nThe tea glass is gone and the bird sits left of the stacked teapot, so both requested actions land. The image also shows an artifact: a conspicuous dark vertical area remains on the teapot body. This is the hardest test because it asks for deletion and spatial repositioning together. For cleaner results, split the workflow into two runs: remove the glass first, then move the bird in the accepted image.<\/p>\n
When to pick Flux Kontext Max Multi<\/h2>\n
\n\n\nNeed<\/th>\n What these outputs support<\/th>\n Best prompt move<\/th>\n<\/tr>\n<\/thead>\n \n\nQuick merch or logo concepts<\/td>\n Good local logo integration, especially on stylized clothing<\/td>\n Name location and approximate scale; inspect the spelling<\/td>\n<\/tr>\n \nInterior mood exploration<\/td>\n Strong scene preservation, looser material fidelity<\/td>\n Describe the surface and accept variation<\/td>\n<\/tr>\n \nStrict portrait retouching<\/td>\n Risky when identity must remain fixed<\/td>\n Protect face, pose, wardrobe, and background explicitly<\/td>\n<\/tr>\n \nMulti-step object changes<\/td>\n Possible, but artifacts can appear<\/td>\n Split removal and relocation into separate runs<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\nChoose Flux Kontext Max Multi when the second image is a visual reference and the output can tolerate interpretation. It is especially useful for rapid apparel, brand-placement, and scene-direction concepts. Do not treat it as pixel-accurate compositing: the hair test and wallpaper test show that the model can reinterpret more than the requested region. Start with a short instruction, lock down anything that must not change, and use separate passes for complex edits.<\/p>\n
Try the model<\/h2>\n
Model and test setup<\/h2>\n
The model page is Flux Kontext Max Multi on Wiro<\/a>. Each test passed two image URLs: the first was the base scene and the second was the reference. Prompt upsampling was set to false<\/strong>, safety tolerance to 2<\/strong>, seed to 42<\/strong>, and output format to JPEG<\/strong>. Keeping prompt upsampling off matters here: the results reflect the direct instructions shown in each caption rather than an expanded prompt.<\/p>\n Wiro’s model documentation shows a completed example with 6.0 seconds<\/strong> elapsed time. It does not publish a fixed per-output price. A separate cancelled-task example lists a total cost of 0.003510<\/strong>, but that is not evidence of a delivered-output price, so it should not be used to budget a production run. Queue time, inputs, and platform settings can also change the elapsed time.<\/p>\n For background on the FLUX family and its API concepts, see the Black Forest Labs documentation<\/a>. For adjacent editing workflows, compare this test with Qwen Image Edit Fast<\/a>, Seedream V5 Lite image-editing tests<\/a>, and GPT Image 1.5 text-edit tests<\/a>.<\/p>\n The output places a large illustrated koala-and-bottle mark on the right side of the white shirt. The man, pose, shirt buttons, and shop-like background remain recognizable. The edit is visually clean, though the result is larger than the prompt’s \u201csmall size\u201d request. This is a useful reminder that placement and scale need separate, explicit constraints when brand artwork matters.<\/p>\n This output has short, swept brown hair, but it also changes the subject’s face, outfit, and framing. It succeeds as a loose visual transfer, not as a strict retouch. Pick this approach for mood boards or concept exploration. For a client headshot, state the protected areas directly: keep face, identity, clothes, pose, and background unchanged; change only hair color and style.<\/p>\n The bed, windows, floor, and wall position survive the edit. The rear wall becomes a dark, aged brick surface rather than a close reproduction of the supplied pattern. That makes the result convincing as an interior redesign concept, but weak for a product-accurate wallpaper preview. Use it to explore surface direction; use a compositing workflow when a specific repeat pattern must remain exact.<\/p>\n The output keeps the selfie-like illustration and puts a readable green \u201cWIRO\u201d wordmark across the shirt. It is centered and local, although again not especially small. Among these outputs, this is the clearest evidence that the model can carry a simple brand mark into a stylized scene without turning it into gibberish.<\/p>\n The model produces a black hoodie with a tidy koala patch near the left chest\/shoulder area. The patch itself looks integrated, but the person and clothing change from the starting portrait. This is a good direction test for merch mockups, not proof that the model will preserve a supplied talent photo. In production, run one change at a time and reject any result that alters the protected subject.<\/p>\n The tea glass is gone and the bird sits left of the stacked teapot, so both requested actions land. The image also shows an artifact: a conspicuous dark vertical area remains on the teapot body. This is the hardest test because it asks for deletion and spatial repositioning together. For cleaner results, split the workflow into two runs: remove the glass first, then move the bird in the accepted image.<\/p>\n Choose Flux Kontext Max Multi when the second image is a visual reference and the output can tolerate interpretation. It is especially useful for rapid apparel, brand-placement, and scene-direction concepts. Do not treat it as pixel-accurate compositing: the hair test and wallpaper test show that the model can reinterpret more than the requested region. Start with a short instruction, lock down anything that must not change, and use separate passes for complex edits.<\/p>\nWhat the six outputs actually show<\/h2>\n
1. Logo on a shirt: reference fidelity with a photographic base<\/h3>\n
2. Hair-look transfer: the edit changes more than hair<\/h3>\n
3. Wall pattern transfer: geometry holds, reference texture does not<\/h3>\n
4. Logo on an illustrated shirt: strong text and local placement<\/h3>\n
5. Sleeve patch: localized edit with a changed subject<\/h3>\n
6. Remove and move: two instructions complete with a visible compromise<\/h3>\n
When to pick Flux Kontext Max Multi<\/h2>\n
\n\n
\n \nNeed<\/th>\n What these outputs support<\/th>\n Best prompt move<\/th>\n<\/tr>\n<\/thead>\n \n Quick merch or logo concepts<\/td>\n Good local logo integration, especially on stylized clothing<\/td>\n Name location and approximate scale; inspect the spelling<\/td>\n<\/tr>\n \n Interior mood exploration<\/td>\n Strong scene preservation, looser material fidelity<\/td>\n Describe the surface and accept variation<\/td>\n<\/tr>\n \n Strict portrait retouching<\/td>\n Risky when identity must remain fixed<\/td>\n Protect face, pose, wardrobe, and background explicitly<\/td>\n<\/tr>\n \n Multi-step object changes<\/td>\n Possible, but artifacts can appear<\/td>\n Split removal and relocation into separate runs<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n Try the model<\/h2>\n