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 2509
#32 of 32 in Image Editing
Seedream 4.0
#15 of 62 in Text-to-Image
Where the votes landed
Qwen Image Edit 2509
0.0%
win rate
Ties
0.0%
Seedream 4.0
100.0%
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 2509
- + Excellent sense of motion and speed with directional blur.
- + High-quality scenic background that perfectly matches the 'California coastline' prompt.
- − The man's clothing from the source image is completely lost.
- − The man is positioned incorrectly behind the back seat rather than in the driver's seat.
Seedream 4.0
- + Successfully preserved the man's identity, including his specific hairstyle, scarf, and plaid coat.
- + Excellent placement of the man in the driver's seat with hands on the wheel.
- + Maintains very high structural consistency with the source car.
- − The motion blur on the wheels is slightly less realistic than Model A's.
Verdict: Seedream 4.0 is the clear winner as it successfully preserved the specific person and his clothing from the second source image while placing him correctly in the driver's seat. Qwen Image Edit 2509 failed significantly by placing the man behind the seats in a physically impossible position and losing his distinctive attire.
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 2509
- + Excellent preservation of the source image's pose.
- + Clean visual style that matches the studio lighting.
- − Completely failed to use Image 2 as a character reference.
- − Did not change the gender, face, or clothing of the person in the source image.
Seedream 4.0
- + Successfully integrated clothing elements from Image 2, including the scarf and black sweatshirt.
- + Made an attempt to alter the face and skin tone to better resemble the character in Image 2.
- + Maintains the yellow background and red ottoman from the source.
- − Modified the pose significantly, losing the cross-legged balance and arm angle of the original.
- − The character's face is highly distorted with severe artifacts and 'double eyes'.
- − The hair from Image 1 was not updated to match the character in Image 2.
Verdict: Both models struggled with this complex multi-image reference task. Qwen Image Edit 2509 essentially ignored the character reference and just slightly sharpened the original image, failing the edit instruction entirely. Seedream 4.0 attempted to merge the character traits (clothing, scarf, skin tone) onto the pose, but the resulting image has severe anatomical distortions, a broken pose, and a corrupted face.
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 2509
- + Successfully applied the full outfit including matching scarf and boots.
- + Kept the facial structure and hair patterns largely consistent with Image 1.
- + Matched the pose well by integrating the person into the scene with full-body context.
- − Included sunglasses and facial hair from Image 2, violating the instruction to keep the face unchanged.
- − Changed the person's body type and skin texture significantly to look more like the model in Image 2.
Seedream 4.0
- + Excellent preservation of the unique vitiligo skin patterns from Image 1.
- + Perfectly replicated the scarf, coat, and jewelry details from Image 2.
- + Maintained the lighting and color balance of the original beach scene.
- − Failed the negative constraint to keep the face unchanged by adding sunglasses and facial hair.
- − The shoe placement and leg positioning look slightly ungrounded compared to the sand.
Verdict: Both models failed the negative constraint to keep the original face exactly 'unchanged' as both added sunglasses and facial hair from the reference outfit. However, Seedream 4.0 was far more successful at preserving the subject's identity by maintaining the specific vitiligo patterns on the face and neck, whereas Qwen Image Edit 2509 replaced the subject's skin and features with those of the model from the second 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 2509
- + Natural-looking hair texture with visible specular highlights consistent with the lighting.
- + Successfully preserved facial features and background without unwanted alterations.
- + Good integration of sideburns with the existing beard.
- − The hair volume is slightly stylized/exaggerated, bordering on a wig-like appearance.
Seedream 4.0
- + Excellent source preservation, maintaining identical background and clothing details.
- + Realistic hair density and volume that feels appropriate for the character.
- − The hairline appears somewhat artificial and lacks the fine transition hairs seen in the other model.
- − The hair texture is a bit fuzzy and lacks the defined strands/lighting of the source beard.
Verdict: Qwen Image Edit 2509 produced a more visually convincing result with hair that better matches the lighting and texture of the original image, despite the style being a bit voluminous. Seedream 4.0 successfully preserved all source details but the resulting hair edit feels flatter and the hairline less natural. Qwen Image Edit 2509 is the winner for its superior rendering of realistic hair strands and integration.
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 2509
- + Successfully incorporates all prompt elements into a cohesive scene.
- + Adopts a traditional cartoon caricature art style with exaggerated features.
- + Preserves the 'selfie' arm gesture from the source image.
- − The text in the speech bubble is nonsensical gibberish.
- − The dog is very small and lacks detail compared to the rest of the image.
- − Completely replaces the source background rather than editing it.
Seedream 4.0
- + Strong preservation of the source background elements while adding the desk.
- + Excellent caricature of the subject's face while retaining the source likeness.
- + High-quality rendering of the dog and hockey equipment.
- − The hand holding the microphone is severely mangled with extra fingers.
- − The arm holding the phone is unnaturally skinny and poorly attached to the shoulder.
- − The transition between the living room background and the news desk is jarring.
Verdict: Qwen Image Edit 2509 creates a more 'complete' caricature by committing fully to an illustrated style and building a full news studio/hockey rink environment. However, Seedream 4.0 is preferred for its superior facial caricature that clearly resembles the user, its realistic dog, and its creative use of the news desk, despite significant anatomical errors in the hands.
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 2509
- + Maintains the original background composition and elements accurately
- + Captures a clean modern anime aesthetic similar to high-budget feature films
- + Excellent preservation of clothing patterns and facial expressions
- − Art style leans more toward generic modern anime than the specific Ghibli painterly look
- − The character in the foreground is significantly sharper than the source image's depth of field
Seedream 4.0
- + Expertly captures the hand-painted watercolor texture typical of Studio Ghibli concept art
- + Highly soft and nostalgic lighting and color palette
- + Character features strongly evoke the Ghibli character design style
- − Loss of background detail, replacing the city street with abstract watercolor washes
- − Loses the specific 'shocked' expression of the girl on the right compared to the source
Verdict: Qwen Image Edit 2509 provides a more faithful recreation of the original photo's composition and details while applying an anime filter, but Seedream 4.0 much more accurately captures the requested 'Studio Ghibli' essence through its watercolor textures and specific character design traits. While Seedream 4.0 loses the background detail, it successfully delivers the 'hand-painted' and 'dreamy' mood requested in the prompt.
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 2509
- + Successfully added both blowing hair and flying leaves as requested.
- + Maintained high image clarity and sharpness.
- + Preserved the details and colors of the original scene very well.
- − The wind effect in the hair looks somewhat artificial and stiff.
- − The leaves appear to be 'pasted' on top rather than integrated into the 3D space.
Seedream 4.0
- + The hair blowing effect looks more natural and soft compared to Model A.
- + Added a warm, sunny lens flare that contributes to the energetic feel.
- + The motion blur on the leaves creates a better sense of dynamic movement.
- − The entire image has a slight blur/softening compared to the original.
- − The dog's face lost some of its sharp detail.
Verdict: Both models followed the instructions well, but Seedream 4.0 captures the 'energetic and lively' feel more effectively through the use of soft motion blur and lighting. While Qwen Image Edit 2509 is sharper, its edits look like static additions rather than captured motion. Seedream 4.0's interpretation of blowing hair is much more realistic and visually appealing.
Explore each model
ByteDance's image generation model with integrated text-to-image and image editing capabilities in a unified architecture, supporting up to 4K resolution