Alibaba's Qwen image editing model for instruction-based image modifications and transformations
Settled by community votes across 6 shared challenges, with an AI judge weighing in on each.
Qwen Image Edit 2511
20.2
arena score
#16 of 32 in Image Editing
Skill signature
· Image Editing
Wan 2.6
23.3
arena score
#28 of 62 in Text-to-Image
Top 2 in Image-to-Video
Vote tally
Where the votes landed
Qwen Image Edit 2511
35.4%
win rate
Ties
2.4%
Wan 2.6
62.2%
win rate
35.4%
2.4% ties
62.2%
Shared challenges
6
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
Man and Car in California
EditingEdit instruction
“Make a photo of the man driving the car down the California coastline”
Source
Qwen Image Edit 2511
Wan 2.6
33% wins
3% ties
64% wins
AI Judge Analysis
Qwen Image Edit 2511
- + Excellent preservation of the subject's face, hair, and clothing.
- + Interior details like the cream seats and wood trim closely match the source car.
- + Realistic integration of the subject into the driver's seat.
- − The car's exterior body lines are slightly warped near the top of the door.
Wan 2.6
- + Captured the California coastline aesthetic very well with motion blur.
- + Successfully combined the subject and the car into a cohesive scene.
- − The subject's facial features and hair texture are significantly altered from the source.
- − The car is positioned for left-hand drive but the steering wheel is on the right side of the cabin (RHD), which is inconsistent with California roads.
Verdict: Qwen Image Edit 2511 is the winner because it maintained incredibly high fidelity to the source images, perfectly preserving the man's face, hairstyle, and specific plaid coat. In contrast, Wan 2.6 lost much of the subject's likeness and introduced a logical error by placing a right-hand drive vehicle on a road meant for right-side driving.
Outfit Transfer Challenge
EditingEdit instruction
“Use Image 1 as the base person. Dress them in the exact elaborate outfit from Image 2 (including all layers, accessories, jewelry, and shoes). Carefully adapt the clothing to the body shape and pose in Image 1 while maintaining realistic fabric behavior, correct proportions, and perfect lighting/shadow matching. Keep the person’s exact face, hair, and background completely unchanged.”
Source
Qwen Image Edit 2511
Wan 2.6
AI Judge Analysis
Qwen Image Edit 2511
- + Successfully transferred the entire outfit including pants and shoes
- + Maintained the full body composition of the scene
- − Completely failed to preserve the person, replacing the original face and skin with the man from Image 2
- − The background structure has noticeable warping and artifacts around the legs
Wan 2.6
- + Perfectly preserved the subject's face, unique skin markings, and sand details
- + Accurately reproduced the textures and pattern of the scarf and coat
- − Cropped the image into a square, losing the lower half of the person and the outfit
- − Added sunglasses that were not requested, though they were present in Image 2
Verdict: Both models struggled with the complexity of the image-to-image transfer. Qwen Image Edit 2511 successfully captured the full outfit but failed the primary constraint by replacing the base person's face and identity entirely. Wan 2.6 did an excellent job of preserving the person's unique physical features but cropped the image, omitting the requested pants and shoes.
Bald man challenge
Image EditingEdit instruction
“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”
Qwen Image Edit 2511
Wan 2.6
75% wins
0% ties
25% wins
AI Judge Analysis
Qwen Image Edit 2511
- + Excellent preservation of the original facial features and glasses
- + Good integration of sideburns with the existing beard
- + Retained the original background and lighting perfectly
- − The hair texture looks slightly oily or plastic-like with unnatural swirls
- − The hairline is a bit high and harsh on the forehead
Wan 2.6
- + The hair texture is extremely realistic and voluminous
- + The hairline and hair integration feel very natural for the subject's age and style
- + Perfectly preserved the jacket and background elements
- − Slightly altered the shape of the glasses and the bridge of the nose
- − The eyes appear subtly different from the source image
Verdict: Both models performed excellent edits with high source preservation. Qwen Image Edit 2511 did a better job of keeping the facial features 100% identical, but the hair itself looks a bit synthetic. Wan 2.6 provided a much more believable and aesthetically pleasing head of hair, though it slightly softened the facial likeness in the process.
Over-the-top cartoon caricature
EditingEdit instruction
“Create a caricature of me and my job. Make it exaggerated and humorous, incorporating my profession as a tv show anchor and my love for dogs and hockey.”
Source
Qwen Image Edit 2511
Wan 2.6
AI Judge Analysis
Qwen Image Edit 2511
- + Excellent capture of the subject's facial features in a caricature style.
- + Strong adherence to the 'tv show anchor' and 'dog' themes.
- + Good use of exaggerated expressions typical of a caricature.
- − Completely missed the hockey element requested in the prompt.
- − Contains garbled non-sensical text in the speech bubble.
- − Includes strange artifacts like a candle on a laptop.
Wan 2.6
- + Successfully incorporated all three prompt elements: tv anchor, dogs, and hockey.
- + The character's face maintains a reasonable likeness to the source image in a cartoon style.
- + Clean, cohesive composition with clear storytelling.
- − The caricature style is more 'generic cartoon' than 'exaggerated caricature'.
- − Minor anatomical issues with how the dog holds the hockey stick.
Verdict: Wan 2.6 is the clear winner because it actually followed the full prompt, including the hockey element which Qwen Image Edit 2511 completely ignored. While Qwen did a slightly better job of creating a 'caricature' rather than a 'cartoon' and captured the specific facial structure of the woman more accurately, its failure to include hockey and its inclusion of gibberish text makes it a less successful edit.
Studio Ghibli Anime Style
EditingEdit instruction
“Transform this photo into a Studio Ghibli–inspired illustration. Use soft pastel colors, hand-painted textures, gentle lighting, dreamy backgrounds, and a warm, nostalgic mood”
Source
Qwen Image Edit 2511
Wan 2.6
67% wins
0% ties
33% wins
AI Judge Analysis
Qwen Image Edit 2511
- + Excellent preservation of the original pose and character likeness.
- + Good balance between a painting style and the original structure.
- + High-quality linework that maintains the expressions of the source image.
- − The style feels more like a modern manhwa/digital illustration than specifically Studio Ghibli.
- − The lighting is a bit flat compared to the 'dreamy' request.
Wan 2.6
- + Successfully captures the hand-painted watercolor texture associated with Ghibli background art.
- + Uses soft pastel colors and atmospheric lighting to create a nostalgic mood.
- + Maintains the composition and character roles perfectly.
- − The white speckles/sparkles feel a bit overlayed and distracting.
- − The character features are slightly more homogenized than in model A.
Verdict: Both models did an exceptional job of preserving the structure of the original 'distracted boyfriend' meme. Qwen Image Edit 2511 provides a cleaner, more precise digital illustration, but Wan 2.6 better captures the specific requested 'Studio Ghibli' aesthetic through its use of soft watercolor textures and gentle, warm lighting.
Golden Hour Stroll
Image EditingEdit instruction
“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”
Qwen Image Edit 2511
Wan 2.6
20% wins
0% ties
80% wins
AI Judge Analysis
Qwen Image Edit 2511
- + Excellent adherence to the 'windy' hair request with strands fanned out symmetrically.
- + Includes a dense concentration of leaves in various colors (orange, yellow, green) for better visual energy.
- + Maintains high fidelity to the original person and dog faces.
- − The hair flow looks a bit artificial and stiffly posed compared to natural movement.
- − Some leaves appear disconnected or lack consistent motion blur.
Wan 2.6
- + Naturalistic hair movement that follows a consistent wind direction.
- + Motion blur on the leaves creates a more realistic sense of movement.
- + Excellent preservation of the original image's lighting and background elements.
- − Fewer leaves added compared to Model A, resulting in slightly less 'lively' energy.
- − The hair edit is more subtle and less transformative than Model A.
Verdict: Both models handled the editing task well, preserving the base image while adding the requested elements. Qwen Image Edit 2511 provided more visual 'pop' by adding a high number of colorful leaves and a very dramatic hair change, while Wan 2.6 opted for a more realistic, subtle approach with consistent wind direction and motion blur. Qwen is slightly preferred for better meeting the request for 'energetic and lively' feel through higher contrast and more numerous additions.
Next steps
Explore each model
Alibaba's multimodal generation model from the Wan AI suite, supporting text-to-video, image-to-video, reference-to-video with audio, and text-to-image, in both Chinese and English