Alibaba's Qwen image editing model for instruction-based image modifications and transformations
Settled by community votes across 7 shared challenges, with an AI judge weighing in on each.
Qwen Image Edit 2511
#16 of 32 in Image Editing
Wan 2.7
#39 of 62 in Text-to-Image
Where the votes landed
Qwen Image Edit 2511
0%
win rate
Ties
0%
Wan 2.7
0%
win rate
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
Man and Car in California
Editing“Make a photo of the man driving the car down the California coastline”
AI Judge Analysis
Qwen Image Edit 2511
- + Excellent preservation of the man's facial features and clothing.
- + Realistic and high-quality interior detailing for the car.
- + Perfect scenery that accurately reflects the California coastline.
- − The car interior/dashboard design was changed from the source image's convertible style.
- − The man's scale relative to the car seems slightly cramped.
Wan 2.7
- + Successfully preserved the original car's exterior design and perspective perfectly.
- + Good composition showing the car moving along the road with motion blur.
- − The man's face and identity are completely lost and replaced with a distorted figure.
- − The person in the car is barely visible and does not represent the man from the source image.
Verdict: Qwen Image Edit 2511 wins this challenge by effectively merging the identity of the man with the requested scene, maintaining his clothing and facial features with high fidelity. While Wan 2.7 did a better job of preserving the exact model of the car from the source image, it failed to realistically include the subject, resulting in a distorted, unrecognizable figure.
Pose & Character Mashup
Editing“Use Image 1 as the exact pose reference and Image 2 as the character reference. Recreate the person/character from Image 2 in the exact dynamic pose and body position from Image 1. Keep the exact face, hair, clothing style/details, and expression from Image 2. Match the lighting and environment of Image 1. The final image must show the character from Image 2 performing the precise action/pose from Image 1 with perfect anatomy and natural integration.”
AI Judge Analysis
Qwen Image Edit 2511
- + Matches the background environment of Image 2.
- + Retains the man from Image 2 in the scene.
- − Fails completely to map the character from Image 2 onto the pose from Image 1.
- − Simply overlays the woman from Image 1 onto Image 2's background.
- − Creates a nonsensical composition with the woman floating in front of the man.
Wan 2.7
- + Perfectly maintains the pose and environment from Image 1.
- − Fails to perform the edit requested.
- − Completely ignores the character reference from Image 2.
- − Provides an output identical to the source Image 1.
Verdict: Both models failed the complex task of character/pose transfer. Qwen Image Edit 2511 attempted to merge elements by layering the woman from Image 1 over the man from Image 2, but it failed to follow the instruction to replace the character. Wan 2.7 simply returned the original pose reference image, ignoring the character reference Image 2 entirely. Qwen is slightly better only because it attempted to integrate elements from both images, whereas Wan 2.7 performed no edit at all.
Outfit Transfer Challenge
Editing“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.”
AI Judge Analysis
Qwen Image Edit 2511
- + Matches the specific clothing items from Image 2 (blue coat, plaid scarf, sunglasses).
- + Successfully integrates black dress shoes into the scene.
- − Completely failed to preserve the person from Image 1, replacing the face and skin characteristics with the person from Image 2.
- − Extensive cropping and movement of the subject changed the composition.
Wan 2.7
- + Perfectly preserved the original person's face, hair, and unique skin features from Image 1.
- + Excellent integration of the pose and body into the beach environment.
- − Hallucinated an entirely different outfit (gold embroidery, multiple necklaces) instead of the one in Image 2.
- − Ignored the specific clothing reference image.
Verdict: This is a case of two different types of failure. Qwen Image Edit 2511 followed the instructions to use the clothing from Image 2 but failed the primary constraint of keeping the person from Image 1 the same. Wan 2.7 perfectly preserved the person from Image 1 and adjusted their pose realistically, but it completely ignored the specific outfit provided in Image 2, creating a generic 'elaborate' costume instead. Wan 2.7 is slightly better as a technical edit because it correctly identifies and preserves the subject's identity as requested, whereas Qwen effectively just performed a face-swap with the reference image.
Bald man challenge
Image Editing“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”
AI Judge Analysis
Qwen Image Edit 2511
- + Excellent preservation of the original facial features and expression.
- + Matches the beard texture well with the new curly hair.
- + Strong adherence to the 'thick head of hair' instruction with believable lighting.
- − The hairline on the side near the ear looks a bit detached or floating.
Wan 2.7
- + Natural, realistic hair volume and style that fits the character's age.
- + Seamless integration of the hair with the original skin and forehead.
- + Perfectly preserves the background and clothing without any artifacts.
- − Slightly alters the shape of the eyebrows and forehead compared to the source.
Verdict: Both models performed exceptionally well on this task, successfully adding hair while maintaining the integrity of the original photo. Wan 2.7 provides a slightly more modern and seamless look with a very natural hairline, whereas Qwen Image Edit 2511 offers a more voluminous curly style that maintains the exact facial structure of the source image more faithfully.
Over-the-top cartoon caricature
Editing“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.”
AI Judge Analysis
Qwen Image Edit 2511
- + Captures a strong TV anchor newsroom aesthetic with a microphone and desk setup.
- + Maintains recognizable facial features from the source image in the caricature style.
- + Includes a cute dog companion as requested.
- − Completely misses the 'hockey' requirement.
- − Includes nonsensical text in the speech bubble.
- − Includes strange artifacts like a candle on the anchor desk.
Wan 2.7
- + Successfully incorporates all three themes: TV anchor, dogs, and hockey (puck, helmet, and rink).
- + Effectively captures the caricature 'big head' style while maintaining a good likeness of the subject.
- + Correctly renders legible English text in the speech bubbles.
- − The composition is a bit cluttered with multiple dogs and overlapping elements.
- − The transition from the 'real' arm into the cartoon body is slightly jarring.
Verdict: Wan 2.7 is the clear winner as it successfully incorporated all three requested elements: the TV anchor profession, dogs, and hockey. Qwen Image Edit 2511 completely ignored the hockey requirement and included gibberish text, whereas Wan 2.7 even cleverly added a hockey helmet to one of the dogs and included a miniature rink.
Studio Ghibli Anime Style
Editing“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”
AI Judge Analysis
Qwen Image Edit 2511
- + Excellent preservation of the original subjects' features and poses.
- + Captures a vibrant, modern anime aesthetic similar to high-budget features.
- + Maintains clear focal points and sharp details on the characters.
- − The 'sparkle' artifacts feel a bit distracting and less like a natural Ghibli texture.
- − The lighting is somewhat generic rather than the specific 'dreamy' quality requested.
Wan 2.7
- + Perfectly captures the hand-painted watercolor texture characteristic of Studio Ghibli backgrounds.
- + The soft, desaturated pastel color palette is highly accurate to the prompt.
- + Atmospheric, gentle lighting creates a more nostalgic mood than the competitor.
- − Slight loss of facial detail compared to the original photo.
- − The line work on the clothing is a bit more simplified.
Verdict: Both models did an excellent job preserving the source image's composition and subjects while applying the requested style. Wan 2.7 is the winner as it much more accurately captures the specific 'hand-painted' and 'watercolor' aesthetic requested in the prompt, whereas Qwen Image Edit 2511 produces a cleaner, more digital-looking anime style.
Golden Hour Stroll
Image Editing“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”
AI Judge Analysis
Qwen Image Edit 2511
- + Successfully added a large number of falling leaves across the scene.
- + Followed the instruction to make the hair blow in the wind.
- + Maintained high image clarity and fidelity to the original background.
- − The blowing hair looks somewhat rigid and unnatural in its symmetry.
- − Some leaves appear to be floating statically rather than having motion blur.
Wan 2.7
- + The hair movement looks more natural and realistic.
- + Preserved the identity of the woman and the dog perfectly.
- + The added leaves feel better integrated into the 3D space of the scene.
- − The 'dynamic motion' effect is a bit subtle compared to Model A.
- − Fewer leaves were added, which slightly reduces the 'energetic' feel requested.
Verdict: Both models followed the instructions well. Qwen Image Edit 2511 provided a more heavy-handed and literal interpretation of the prompt with many leaves and dramatic hair, whereas Wan 2.7 produced a more subtle and visually realistic result. Wan 2.7 is preferred for its more natural-looking hair physics, although Qwen Image Edit 2511 captured the 'energetic' request more overtly.
Explore each model
Alibaba's Wan 2.7 image generation and editing model for text-to-image, reference-guided generation, and instruction-based image edits