Black Forest Labs' open-weights multimodal flow transformer for in-context image generation and editing, available for non-commercial use with character consistency and style transfer capabilities
Settled by community votes across 4 shared challenges, with an AI judge weighing in on each.
FLUX.1 Kontext [dev]
#54 of 62 in Text-to-Image
HiDream E1
#33 of 32 in Image Editing
Where the votes landed
FLUX.1 Kontext [dev]
0%
win rate
Ties
0%
HiDream E1
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
FLUX.1 Kontext [dev]
- + Successfully added a realistic full head of hair that matches the character style.
- + Excellent preservation of the original jacket, lighting, and background.
- + The hair texture and lighting integration on the scalp look natural.
- − Subtly smoothed out some facial wrinkles, losing a bit of the original's rugged texture.
- − Changed the frame color and shape of the glasses slightly.
HiDream E1
- + Followed the instruction to add a full head of hair.
- − Failed to preserve the source image, significantly altering the background and skin texture.
- − The hair has a 'wig-like' artificial quality with unnatural waves.
- − Introduced heavy artifacts and a strange speckled texture over the entire face.
Verdict: FLUX.1 Kontext [dev] is the clear winner as it successfully applied the requested edit while maintaining the integrity of the original photo's composition, lighting, and garment details. HiDream E1 failed significantly by changing the entire background environment and introducing distracting digital artifacts across 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
FLUX.1 Kontext [dev]
- + Maintains the subject's brown hair color from the source image
- + Successfully incorporates the 'tv show anchor' profession via a television display and news-style graphics
- + Good caricature art style that feels playful and bold
- − Completely missed the hockey theme mentioned in the prompt
- − Nonsensical text rendering like 'JOB' and garbled letters on the dog's bowl
HiDream E1
- + Excellent integration of all three themes: anchor (suit/microphone), dogs, and hockey (rink background)
- + High visual quality and polished 3D-render style caricature
- + The composition effectively places the subject in a dynamic environment
- − Changed the subject's hair color from brown to blonde, losing source identity
- − The hands holding the dog have anatomical errors with distorted fingers
Verdict: HiDream E1 followed the complex prompt instructions much better by including all three requested elements (anchor, dog, and hockey), whereas FLUX.1 Kontext [dev] omitted hockey entirely. Although HiDream E1 incorrectly changed the subject's hair color, its 3D caricature style is much more visually engaging and aligned with the 'humorous and exaggerated' requirement.
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
FLUX.1 Kontext [dev]
- + Excellent preservation of the source image's composition and specific clothing details like the plaid shirt pattern.
- + Captures the iconic character expressions from the original meme perfectly.
- + Clean linework and professional anime aesthetic.
- − Looks more like modern digital anime than the specific hand-painted, soft texture style of Studio Ghibli.
- − The lighting is somewhat flat compared to the 'dreamy' request.
HiDream E1
- + Successfully applies the soft pastel color palette and warm, nostalgic mood requested.
- + The background has a softer, more painterly feel consistent with the prompt.
- − Lost significant source detail, such as the plaid pattern on the man's shirt which is now solid pink.
- − The character expressions are much weaker and lose the narrative tension of the original image.
- − Poor hand rendering on the man's right hand.
Verdict: FLUX.1 Kontext [dev] is the clear winner for its exceptional ability to preserve the source image's identity, including the complex plaid pattern and the specific facial expressions that make the meme recognizable. While HiDream E1 captured the requested pastel color palette more accurately, it failed to maintain the essential details of the input image and suffered from anatomical errors in the hands.
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
FLUX.1 Kontext [dev]
- + Excellent source preservation, maintaining the woman's face and the dog's features accurately.
- + Subtle and realistic hair movement that feels integrated with the original lighting.
- + Natural leaf additions that match the environment scale.
- − The 'flying leaves' are very small and sparse, making it less 'energetic' than requested.
HiDream E1
- + Strong adherence to the request for 'energetic' feel with very visible hair movement.
- + Clearly visible leaves that add a sense of autumn and motion.
- − Significant loss of source preservation, changing the woman's facial features and the dog's expression.
- − The added leaves look like flat, artificial overlays without realistic depth or motion blur.
- − The overall image quality has a slightly 'over-processed' look compared to the original.
Verdict: FLUX.1 Kontext [dev] is the clear winner for image editing because it successfully applied the requested changes while perfectly preserving the identity of the person and dog from the source image. In contrast, HiDream E1 failed the preservation task by noticeably altering the faces and adding leaves that look like poorly integrated stickers, despite following the 'energetic' prompt more literally.
Explore each model
HiDream AI's image-to-image editing model for instruction-based image modifications and transformations