GPT Image 1 Mini OpenAI vs P-Image Edit PrunaAI

GPT Image 1 Mini

• + Excellent preservation of the man's identity, clothing, and hairstyle.
• + Perfect integration of the man into the driver's seat with realistic hands on the wheel.
• + Accurate preservation of the specific Rolls Royce details while adding motion blur.

• − The transition between the man's torso and the seatbelt is slightly soft.
• − The scale of the man relative to the car is slightly large.

P-Image Edit

• + Great scenic composition of the California coastline.
• + Good sense of motion and speed.

• − The man is too small and his features are almost completely obscured.
• − Does not preserve the man's specific clothing or hairstyle from the source image.
• − The car's interior details on the passenger side look distorted.

GPT Image 1 Mini

• + Maintains the character identity from Image 2 effectively including facial features and sunglasses.
• + Follows the clothing requirements accurately by applying the scarf and black sweater pattern.
• + The background and ottoman color match the source image well.

• − Failed significantly to replicate the complex, dynamic pose from Image 1, opting for a standard crouch.
• − The left hand is distorted with an incorrect number of fingers.

P-Image Edit

• + Attempts the precise, difficult body positioning and leg crossing from Image 1.
• + Successfully incorporates elements from both images including the red sleeve and the long hair from the pose reference.

• − Poor anatomical coherence with a leg appearing to grow out of the back and nonsensical arm placement.
• − Inconsistent character details, blending the male character's face with the female character's long hair and skin tone.
• − The scarf physics and integration with the body are unrealistic.

GPT Image 1 Mini

• + Excellent preservation of the subject's face, skin markings (vitiligo), and features from Image 1.
• + Accurately replicates the coat, scarf pattern, and watch from Image 2.
• + High visual quality with realistic lighting and shadows integrated into the original scene.

• − Changes the hair style slightly (making him bald) instead of keeping it completely unchanged.
• − The scarf pattern is simplified compared to the source.

P-Image Edit

• + Successfully includes the sunglasses and gold chain from Image 2.
• + Maintains the background of Image 1 well.

• − Completely fails to preserve the person from Image 1, replacing him with the man from Image 2.
• − The hand on the scarf has anatomical distortions.
• − The lighting on the subject feels artificial and pasted-on compared to the background.

GPT Image 1 Mini

• + Successfully added thick, dense hair.
• + Good preservation of the overall lighting and background.

• − Substantially altered the man's facial features and nose shape.
• − The person in the image looks like a different individual than the source.

P-Image Edit

• + Excellent preservation of the source person's unique facial structure and features.
• + Near-perfect source preservation of the background and clothing.
• + Realistic curly texture that blends well with the beard.

• − The hairline on the forehead looks slightly superimposed compared to the original skin texture.

GPT Image 1 Mini

• + Successfully captures the likeness of the source subject in a traditional caricature style.
• + Clearly incorporates all requested elements: news anchor desk, microphone, dog, hockey stick, and puck.
• + The hand drawing the microphone is rendered correctly with five fingers.

• − The 'NEWS' text and overall background are quite simple compared to the other model.

P-Image Edit

• + Exuberant composition that cleverly combines a TV studio desk with a hockey rink background.
• + Maintains a very strong facial likeness to the source image while exaggerating features for the caricature.
• + Highly detailed environment with many humorous elements related to the prompt.

• − The dog is very small and has a strange, red mechanical-looking growth on its back.
• − The hand holding the microphone has six fingers and some anatomical distortion.
• − The text on the background screens is nonsensical AI gibberish.

GPT Image 1 Mini

• + Successfully captures a hand-drawn, colored-pencil texture characteristic of specific Ghibli concept art.
• + Preserves the overall layout and warm color palette requested.
• + Maintains the distinct facial expressions of all three characters from the source meme.

• − The textures look more like crayon or chalk than the crisp cel-shaded look typical of Ghibli films.
• − Lack of high-contrast lighting makes the image look somewhat flat.

P-Image Edit

• + Excellent adherence to Ghibli's classic cel-shaded animation style.
• + Successfully replaces the urban background with a 'dreamy' natural landscape as requested.
• + Preserves the identity and clothing of the original subjects with high fidelity.

• − The woman in the foreground has a face that looks a bit too modern/western compared to Ghibli's simplified character designs.
• − Minor anatomical oddity where the man's arm meets his hand near the bottom center.

GPT Image 1 Mini

• + Excellent adherence to the 'flying leaves' instruction with good scale and variety.
• + Successfully modified the hair to flow naturally in a breeze.
• + Maintained high visual consistency with the subjects' faces and clothing.

• − The leaf placement looks slightly like a flat overlay in some areas.
• − Lost some of the background bridge details.

P-Image Edit

• + Impressive preservation of the original source image's geometry and background.
• + High-energy hair movement that feels very dynamic as requested.
• + Leaves have varied colors that add to the lively feel.

• − The leaves are very small and look like particles rather than foliage.
• − Some strange artifacts at the tips of the hair where it meets the sky.