Black Forest Labs' compact, open-source image generation model with sub-second inference, optimized for production and near real-time applications with multi-reference support
Settled by community votes across 7 shared challenges, with an AI judge weighing in on each.
FLUX.2 [klein] 4B
#32 of 62 in Text-to-Image
P-Image Edit
#29 of 32 in Image Editing
Where the votes landed
FLUX.2 [klein] 4B
0%
win rate
Ties
0%
P-Image Edit
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
FLUX.2 [klein] 4B
- + Excellent preservation of the car's original details and lighting.
- + The subject's face and clothing (plaid coat) are recognizably carried over from the source image.
- + Realistic integration of the car onto the road with motion blur on the wheels.
- − The driver's position is slightly awkward, appearing a bit high in the seat for the car's scale.
- − The background lighting is a bit flat compared to the bright highlights on the car.
P-Image Edit
- + Great composition showing more of the winding coastline and classic California scenery.
- + Excellent application of motion blur to the road and wheels to indicate speed.
- + Successfully captures the specific hairstyles of the subject from the source image.
- − The driver's face is obscured by shadows and less recognizable than Model A.
- − The car has some minor distortions, particularly the height of the front grill and the angle of the windshield.
- − The lighting on the car doesn't perfectly match the golden/afternoon light of the landscape.
Verdict: Both models handled this complex image-blending task remarkably well. FLUX.2 [klein] 4B is the winner because it maintained a higher level of detail and likeness for both the car and the man, including his specific plaid coat. While P-Image Edit created a more dynamic landscape, it lost the facial features of the subject in the process.
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
FLUX.2 [klein] 4B
- + Matches the physical pose and body orientation of Image 1 almost perfectly.
- + Effectively combines the character's facial features and clothing items like the scarf and black sweater.
- + Maintains the lighting and environment colors with high fidelity.
- − Conflates the hair from Image 1 and Image 2 into a messy, feminine length that doesn't match the male character.
- − The face appears slightly distorted and over-smoothed compared to the source.
P-Image Edit
- + Successfully captures the male character's short haircut from Image 2 while attempting the pose.
- + Correctly integrates the sunglasses and scarf into the composition.
- − Fails to follow the exact pose from Image 1, significantly altering the leg and arm positions.
- − Serious anatomical issues, including a floating third leg/foot and a hand emerging from the hair.
- − Inconsistent clothing colors, mixing the red from Image 1 with the black from Image 2 on the same garment.
Verdict: FLUX.2 [klein] 4B is the clear winner as it successfully replicates the complex, dynamic pose of Image 1 while incorporating the character details from Image 2. P-Image Edit fails the core request by changing the pose and introducing severe anatomical artifacts, such as extra limbs and misplaced body parts.
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
FLUX.2 [klein] 4B
- + Excellent preservation of the subject's face, skin patterns, and hair
- + High detail in the scarf texture and jewelry
- + Consistent lighting and high-quality lens effects that match the original photograph
- − Crop is slightly tighter than the original
- − Missing the lower half of the body requested (shoes) due to framing
P-Image Edit
- + Successfully includes the full outfit including the shoes
- + Maintains the subject's pose and background context well
- − The subject's facial features and skin patterns have been noticeably altered
- − The fabric of the coat looks unnaturally stiff and flat
- − Shoes do not match the style of the clothing in the reference image
Verdict: FLUX.2 [klein] 4B is the clear winner for its superior identity preservation and high visual quality, although it failed to render the full body. P-Image Edit captured the full attire including shoes, but severely degraded the person's face and vitiligo patterns, failing the 'keep person's exact face unchanged' requirement.
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.2 [klein] 4B
- + Excellent integration with the existing beard and sideburns
- + Natural-looking hairline and texture
- + Perfectly preserves original facial features and environment
- − Slightly messy flyaway hairs on the top
P-Image Edit
- + Successfully added a large volume of hair
- + Maintains original facial structure and background
- − Hair texture looks slightly plastic or wig-like
- − The hairline is unnaturally high and straight
- − Hair color and texture do not match the existing beard as well as Image A
Verdict: FLUX.2 [klein] 4B followed the prompt with superior realism, creating a hairstyle that looks like it naturally belongs to the subject and matches the beard. P-Image Edit provides a full head of hair, but the texture is a bit repetitive and the transition at the forehead is less convincing.
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.2 [klein] 4B
- + Successfully captures the TV anchor profession with a news desk and studio monitors.
- + Excellent use of dogs as co-anchors in a humorous way.
- + Maintains a consistent caricature style throughout the illustration.
- − Completely misses the 'hockey' requirement.
- − Subject's face slightly loses the likeness of the source image in the caricature translation.
P-Image Edit
- + Incorporates all three elements: TV anchor (microphone/desk), dogs (small mascot), and hockey (background and rink).
- + Maintains better facial likeness to the source person in the caricature style.
- + Retains the denim shirt from the original image while modifying the context.
- − The hockey players in the background look somewhat detached from the central figure's scale.
- − The 'dogs' element is less central and impactful compared to Model A.
Verdict: While FLUX.2 [klein] 4B created a more cohesive studio setting and better integrated the dogs, it failed to include the requested hockey theme. P-Image Edit successfully incorporated all three specific prompt requirements (TV, dogs, hockey) while doing a better job of preserving the subject's facial features and clothing style from the source image.
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.2 [klein] 4B
- + Excellent adherence to the 'Studio Ghibli' aesthetic with soft watercolor textures and line work.
- + Successfully transforms the background into a dreamy, hand-painted village scene while keeping the character placement.
- + Captures the requested soft pastel color palette and warm mood perfectly.
- − The facial expression of the girlfriend on the right has lost its original indignant/angry character and is now too calm.
- − The man's neck and jawline are slightly distorted compared to the source.
P-Image Edit
- + Preserves the original source character expressions and likenesses very accurately.
- + Maintains the composition and specific details of the clothing and hair better than Model A.
- + Includes a vibrant, illustrative background that follows the prompt.
- − The style looks more like a standard digital manhwa or generic illustration than the specific Studio Ghibli style requested.
- − The lighting lacks the 'gentle' and 'nostalgic' quality found in Model A.
Verdict: FLUX.2 [klein] 4B followed the stylistic instructions much better, creating an image that genuinely feels like a Ghibli watercolor painting, though it sacrificed the specific 'jealous' expression of the character on the right. P-Image Edit was more faithful to the source characters' faces and expressions, but its artistic style is more generic digital art and failed to capture the hand-painted, soft aesthetic requested by 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
FLUX.2 [klein] 4B
- + Excellent preservation of the original person face and dog's features.
- + Leaves look natural and integrated into the scene.
- + Hair movement is realistic and follows a consistent wind direction.
- − The leash handle area has small background inconsistencies.
P-Image Edit
- + Dramatic hair movement that clearly conveys a strong wind.
- + Good preservation of the overall background scene.
- − The leaves appear like flat, bright stickers rather than organic elements.
- − The hair has a slightly 'stringy' and digital appearance compared to the source.
Verdict: FLUX.2 [klein] 4B is the clear winner as it successfully adds the requested motion while maintaining the high-quality realism of the original image. P-Image Edit creates more dramatic movement in the hair, but the added leaves look like low-quality clip-art overlaid on the scene, which breaks the immersion.
Explore each model
PrunaAI's sub-1-second multi-image editing model supporting up to 5 reference images with state-of-the-art quality