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 4.5
#9 of 62 in Text-to-Image
Where the votes landed
Qwen Image Edit 2509
0%
win rate
Ties
0%
Seedream 4.5
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 preservation of the car's exterior design and badges.
- + Dynamic lighting and sense of motion via wheel blur.
- + Accurate and beautiful depiction of the California coastline.
- − The man's features and clothing are barely identifiable compared to the source.
- − The man is positioned too far back/small in the frame.
Seedream 4.5
- + Exceptional preservation of the man's identity, including his specific plaid coat, scarf, and hairstyle.
- + Perfectly captures the 'driving the car' instruction with clear interior detail.
- + Maintains the specific interior color and trim of the source car.
- − Minor distortion of the car's proportions in the close-up crop.
- − The car door is inexplicably open while the car appears to be in motion.
Verdict: Seedream 4.5 is the clear winner for its incredible ability to preserve the specific clothing and identity of the man from the source image, even managing to keep his unique hairstyle and accessories intact. While Qwen Image Edit 2509 creates a more cinematic wide shot of the car, it essentially replaces the person with a generic figure, whereas Seedream 4.5 successfully merges the two source images into a believable scene.
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
- + Maintains the exact colors and lighting of the source pose image.
- − Completely failed the edit by providing an almost identical copy of the pose image without the character from Image 2.
- − Does not include the face/character, clothing style, or accessories requested in the prompt.
Seedream 4.5
- + Successfully integrated the character from Image 2 into the pose from Image 1.
- + Accurately reproduced the character's clothing, including the scarf and sunglasses.
- + Captured the facial features and expression remarkably well.
- − The anatomy of the right arm (pointing up) is slightly elongated and distorted at the wrist.
- − The transition between the bare feet and the black trousers is somewhat abrupt.
Verdict: Qwen Image Edit 2509 entirely failed the task, producing a slight modification of the source pose image with none of the requested character features. Seedream 4.5 successfully followed all instructions, placing the character from Image 2 into the complex pose and environment of Image 1 while maintaining consistent clothing and accessories.
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 replicates the full-body pose and environment of the person in the source image.
- + Correctly identifies and places all items from Image 2, including the shoes and full jeans.
- − Failed to preserve the face of the original person, blending it with the person from Image 2.
- − Visual quality is low with significant AI artifacts and 'clay-like' skin textures.
Seedream 4.5
- + Perfectly preserves the face and hair of the original person as requested.
- + Maintains high resolution and realistic skin textures, including the vitiligo patterns.
- − Failed to adapt the clothing to the original person's pose, resulting in the person's torso being visible through the shirt.
- − The scarf and coat are composited awkwardly over the bare chest.
Verdict: Seedream 4.5 is the preferred model because it adhered to the strict instruction to keep the person's face and hair completely unchanged, whereas Qwen Image Edit 2509 modified the facial features significantly. Although Seedream 4.5 struggled with the logic of how the coat covers the torso, its visual quality and identity preservation are vastly superior.
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
- + Successfully added a very thick and voluminous head of hair.
- + Maintained the color palette and background of the original image.
- + The hair texture clearly matches the lighting of the scene.
- − The hair looks somehwat like a wig and sits awkwardly on top of the head.
- − The hairline transitions are a bit harsh around the temples.
Seedream 4.5
- + Natural-looking hair texture that blends seamlessly with the man's existing beard and sideburns.
- + Highly realistic hairline and head shape preservation.
- + Excellent preservation of the facial features and lighting from the source image.
- − The hair is somewhat less 'full' and 'thick' compared to the dramatic change in Model A.
Verdict: While Qwen Image Edit 2509 provided more volume, Seedream 4.5 produced a significantly more believable and realistic result. Seedream 4.5 managed to integrate the new hair with the existing sideburns and skin texture in a way that looks like a genuine transformation, whereas Qwen's result appears slightly artificial.
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 adopts a classic comic-book caricature style.
- + Cleverly merges the concept by placing the news desk in front of a hockey rink.
- + Preserves the subject's outfit and distinctive selfie pose from the original image.
- − Text in speech bubbles and on screens is nonsensical.
- − The 'dog' addition is very small and lacks detail compared to the rest of the image.
Seedream 4.5
- + Excellent high-resolution rendition of the subject with 'bobblehead' caricature proportions.
- + Incorporates all elements (mic, hockey sticks, dog, script) with high visual fidelity.
- + Maintains strong facial likeness to the source person.
- − The transition from the source background to the news desk is a bit messy with furniture overlapping.
- − The hands are significantly smaller and less detailed than the rest of the figure.
Verdict: Qwen Image Edit 2509 creates a more creative and integrated scene by putting the news desk directly on ice, though it suffers from typical AI text issues. Seedream 4.5 provides a much higher quality, polished '3D' caricature look that captures the subject's likeness better and arranges the required props more clearly on a studio desk, making it the more visually appealing choice.
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 high fidelity to the source character features despite the stylistic change.
- + Excellent preservation of the background architecture and the red bus detail.
- + Effective use of warm, nostalgic lighting across the composition.
- − Style leans more towards modern digital manhwa/webtoon than the hand-painted Studio Ghibli aesthetic.
- − Changed the iconic red dress to pink.
Seedream 4.5
- + Perfectly captures the Studio Ghibli aesthetic with watercolor textures and iconic character art styles.
- + Preserves the original color palette including the vibrant red dress.
- + Excellent hand-painted feel in the background scenery.
- − Loss of some specific facial likeness from the original models.
- − The hair rendering on the woman in the foreground is slightly simplified.
Verdict: While both models successfully translated the meme into an illustration, Seedream 4.5 captures the specific 'Studio Ghibli' request far better through its use of watercolor textures and classic anime facial structures. Qwen Image Edit 2509 maintains better source preservation in the background and character likeness, but its style feels more like a generic modern digital illustration rather than the requested hand-painted Ghibli look.
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
- + Excellent preservation of the woman's face and original clothing details
- + Clearly defined 'flying leaves' added throughout the frame
- + Hair motion is significant and meets the prompt's request
- − The wind direction is inconsistent between the hair (blowing left) and the dog's tail (blowing right)
- − The added hair strands have a slightly mechanical, repetitive wavy pattern
Seedream 4.5
- + Natural-looking hair motion that blends well with the existing hairstyle
- + Successfully captures a more 'energetic' feel by slightly adjusting the woman's pose
- + Leaves have natural motion blur and depth, enhancing the sense of movement
- − Slightly alters the woman's facial features and expression compared to the original
- − Minor loss of detail in the background foliage compared to the source image
Verdict: Qwen Image Edit 2509 does a better job of preserving the exact identity and details of the subject from the source image, though the wind effects feel a bit artificial. Seedream 4.5 creates a more cohesive and lively atmosphere with realistic motion blur on the leaves and more natural hair movement, but it slightly modifies the subject's face in the process. Qwen Image Edit 2509 is the winner for its superior adherence to the 'source preservation' aspect of the editing task.
Explore each model
ByteDance's latest image generation model unifying text-to-image and image editing in a single architecture, with improved text rendering and 30-40% faster generation than v4.0