HiDream AI's image-to-image editing model for instruction-based image modifications and transformations
Settled by community votes across 4 shared challenges, with an AI judge weighing in on each.
HiDream E1
#33 of 32 in Image Editing
Wan 2.5 (Preview)
#27 of 62 in Text-to-Image
Where the votes landed
HiDream E1
0%
win rate
Ties
0%
Wan 2.5 (Preview)
0%
win rate
Challenge by challenge
The strongest take from each model on every shared challenge, with the AI judge's read.
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
HiDream E1
- + Fulfilled the request for a thick head of hair.
- + Maintained the beard and jacket structure.
- − Failed to preserve the background, changing the landscape to a different desert style.
- − Significantly altered the facial texture, adding unusual skin marks.
- − Hair texture looks synthetic and lacks a natural hairline.
Wan 2.5 (Preview)
- + Excellent source preservation, keeping the background and clothing identical to the original.
- + The hair looks highly realistic with natural texture and a convincing hairline.
- + Perfectly preserved facial features and lighting from the source image.
- − None notable; the edit is seamless.
Verdict: Wan 2.5 (Preview) provided an exceptional edit that seamlessly integrated realistic hair while preserving every other detail of the source image. In contrast, HiDream E1 failed significantly at source preservation by changing the background landscape and adding artifacts to the subject's face.
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
HiDream E1
- + Features a 3D-shaded caricature style that maintains some facial resemblance to the source.
- + Strongly incorporates the hockey theme with a rink background and players.
- + The dog is well-integrated into the central composition.
- − Significant anatomy issues with several extra fingers on both of the character's hands.
- − Faces of the hockey players in the background are distorted and uncanny.
- − The career aspect is less clear than Model B, as she looks more like a fan than an anchor.
Wan 2.5 (Preview)
- + Excellent adherence to the 'TV anchor' job description with a desk, microphone, and monitor.
- + Captures a classic 2D caricature style with better exaggerated proportions.
- + Maintains the outfit color scheme from the source image (denim shirt and black top).
- − The facial features are more generic and lose some of the specific identity of the source subject.
- − The hockey stick is positioned awkwardly, seemingly floating in front of the desk.
- − The background within the TV monitor shows a field rather than an ice rink.
Verdict: Wan 2.5 (Preview) is the superior model for this edit because it successfully incorporates all three requested themes—the TV anchor job, dogs, and hockey—in a clear, humorous composition. While HiDream E1 creates a visually striking 3D render, it suffers from severe hand artifacts and fails to clearly represent the 'TV anchor' profession, making it look more like a general sports fan.
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
HiDream E1
- + Successfully adopts a warm, pastel color palette
- + Captures the basic composition and silhouettes of the original
- + Creates a clean, simplified cartoon style
- − Loses character details like the plaid pattern on the shirt
- − The art style is generic web-comic rather than Ghibli-inspired
- − Changed the woman's red dress to a pale peach color
Wan 2.5 (Preview)
- + Excellent adherence to the Studio Ghibli art style with authentic character designs
- + High preservation of original details like the plaid shirt and specific facial expressions
- + Masterful use of soft lighting, hand-painted textures, and atmospheric floating leaves
- − None notable; the interpretation perfectly matches the prompt and source image
Verdict: Wan 2.5 (Preview) produced a superior result by perfectly capturing the specific aesthetic of Studio Ghibli while preserving the composition and intricate details of the source image, such as the plaid shirt. HiDream E1 opted for a more generic cartoon style and lost significant color and pattern information from the original photo.
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
HiDream E1
- + Successfully applied the wind effect to the hair with clear directional flow.
- + Preserved the identity and pose of the subjects well.
- + The added leaves create a bold sense of movement.
- − The red leaves look like flat, artificial clip-art rather than part of the 3D space.
- − Hair texture became slightly more painterly and lost the realism of the source image.
Wan 2.5 (Preview)
- + Excellent preservation of the original image's lighting and color palette.
- + Naturalistic modification of the hair that integrates seamlessly with the original texture.
- + Added leaves are subtle and feel more geographically appropriate for the summer park setting.
- − The 'energetic and lively' feel is slightly less intense than the other model.
Verdict: Both models followed the instructions well. HiDream E1 provided a more dramatic transformation with higher energy, but the added leaves look very artificial. Wan 2.5 (Preview) is the winner because it successfully added wind-blown hair and motion while perfectly maintaining the photographic quality and lighting of the source image.
Explore each model
Alibaba's text-to-image and image-to-image generation model from the Wan AI suite, offering high-quality visual generation capabilities