FLUX.2 [dev] Black Forest Labs HiDream E1 HiDream AI

Settled by community votes across 4 shared challenges, with an AI judge weighing in on each.

FLUX.2 [dev]

24.5 arena score

#18 of 62 in Text-to-Image

Skill signature · Text-to-Image

HiDream E1

0.7 arena score

#33 of 32 in Image Editing

Vote tally

Where the votes landed

FLUX.2 [dev]

win rate

Ties

HiDream E1

win rate

Shared challenges 4

Challenge by challenge

The strongest take from each model on every shared challenge, with the AI judge's read.

Bald man challenge

Image Editing

Edit instruction

“Give the person a full, thick head of natural hair with realistic texture, density, and a natural hairline. Preserve facial features and lighting.”

FLUX.2 [dev]

HiDream E1

AI Judge Analysis

FLUX.2 [dev]

+ Excellent preservation of the source image pixels outside of the hair area.
+ Extremely realistic hair texture and density that integrated well with the existing beard.
+ Maintains original lighting and facial structure perfectly.

− The choice of an afro style, while fulfilling 'full and thick', might be a narrow interpretation of the prompt.

HiDream E1

+ Successfully adds a full head of hair that fits a common masculine style.

− Completely changes the background environment from a scrubland to a sandy desert.
− Heavily alters the man's facial features and adds significant skin texture artifacts.
− The hair texture looks artificial and doll-like, failing the 'realistic texture' requirement.

Verdict: FLUX.2 [dev] followed the editing instructions perfectly, adding a realistic head of hair while keeping the background, facial features, and lighting identical to the source. HiDream E1 failed as an image editor, effectively regenerating a new image that changed the background, clothing details, and the man's face, while also introducing a plastic hair texture.

Over-the-top cartoon caricature

Editing

Edit instruction

“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.”

Source

FLUX.2 [dev]

HiDream E1

AI Judge Analysis

FLUX.2 [dev]

+ Expertly captures the subject's facial features in a classic caricature style.
+ Perfectly integrates all requested themes including hockey rinks, dogs, and a news anchor desk.
+ Maintains a high level of detail and artistic coherence throughout the composition.

HiDream E1

+ Successfully adopts a 3D animation/caricature style.
+ Includes all requested elements: hockey players, a dog, and professional attire.

− Loses the specific facial likeness of the source image completely.
− Suffers from significant anatomy errors, particularly with the dog's legs and the woman's hands.
− The composition feels less balanced and slightly cluttered in the background.

Verdict: FLUX.2 [dev] is the clear winner as it successfully translates the specific facial features of the woman from the source image into an exaggerated caricature, whereas HiDream E1 generates a generic character that looks nothing like the original. FLUX.2 [dev] also demonstrates superior creativity by placing the dogs in hockey jerseys and integrating the news desk directly onto the ice, while HiDream E1 has several anatomical glitches with the dog and hands.

Studio Ghibli Anime Style

Editing

Edit instruction

“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”

Source

FLUX.2 [dev]

HiDream E1

AI Judge Analysis

FLUX.2 [dev]

+ Excellent character preservation maintaining the specific facial features and expressions of the source image.
+ Perfectly captures the Studio Ghibli aesthetic with soft watercolor textures and a dreamy flower field background.
+ Maintains complex details like the plaid pattern on the man's shirt while translating them into an illustrated style.

− Changes the red dress to a light pink, although this fits the pastel prompt.

HiDream E1

+ Strongly adopts a simplified anime art style.
+ Preserves the general layout and color scheme of the original urban environment.

− Loses the nuances of the characters' expressions, particularly the jealousy/disbelief of the woman on the right.
− The shirt pattern is completely lost, becoming a solid pink garment.
− The art style feels more like generic modern western web-animation than the specific Ghibli aesthetic requested.

Verdict: FLUX.2 [dev] is the clear winner as it successfully translates the source image into the specific Studio Ghibli aesthetic while meticulously preserving the distinct facial expressions and clothing details of the original meme. HiDream E1 creates a much more generic illustration that loses the character depth and the 'hand-painted' texture requested in the prompt.

Golden Hour Stroll

Image Editing

Edit instruction

“Add dynamic motion to this photo: make hair blow in the wind, add leaves flying, energetic and lively feel.”

FLUX.2 [dev]

HiDream E1

AI Judge Analysis

FLUX.2 [dev]

+ Excellent source preservation, maintaining the woman's face and dog's appearance perfectly.
+ Realistic wind effect on hair that flows naturally from the scalp.
+ Subtle and believable integration of leaves with varied motion blur.

− The leash handle in the woman's hand slightly loses its original shape.

HiDream E1

+ Strongly communicates the idea of wind through dramatic hair spread.
+ Large, vibrant leaves make the edit very obvious at a glance.

− Poor source preservation; modifies the woman's facial features and the dog's expression significantly.
− The leaves are unnaturally large and lack realistic textures, looking like clip-art overlays.
− Low-quality resolution compared to the original, with visible smoothing and loss of detail.

Verdict: FLUX.2 [dev] is the clear winner as it successfully adds the requested dynamic motion while perfectly preserving the identity of the person and dog from the source image. HiDream E1 fails as an image editing tool because it significantly alters the facial features of the woman and the details of the dog, while adding leaves that lack realistic integration.