Seedream 4.5 ByteDance vs Vidu Q2 ShengShu Technology

Seedream 4.5

• + Excellent photorealistic lighting and textures
• + Perfect execution of the plant visible through the glass
• + Very clean glass materiality and refractions

• − The sphere is quite small relative to the space

Vidu Q2

• + Strong colors and sharp detail
• + Accurate placement of all requested elements
• + Detailed texture on the book cover

• − The glass refraction of the plant is less convincing than Model A
• − The shadows on the table are a bit harsh and busy

Seedream 4.5

• + Excellent preservation of the subject's clothing and appearance from the source image.
• + Realistic interaction between the man and the car interior.
• + Maintains the specific car model details and interior color.

• − The car door is awkwardly open while ostensibly driving.
• − The composition is a bit tight, cutting off much of the car.

Vidu Q2

• + Provides a more dynamic composition showing the full car and the classic coastline road.
• + Correctly places the driver in a closed vehicle for a driving shot.
• + Good environment lighting and reflections on the car body.

• − The subject's face and hair are significantly altered and lack detail compared to the source.
• − The man's outfit from the source image (plaid coat, scarf) is mostly lost or simplified.

Seedream 4.5

• + Excellent execution of motion blur with the passing car.
• + Authentic atmosphere with realistic wet pavement reflections and raindrops on the raincoat.
• + Strong adherence to the 'imperfect framing' and 'candid' street photography style.

• − The bike chain and gear mechanics are slightly simplified and physically incorrect.

Vidu Q2

• + High level of detail in the bicycle's mechanical components and weathered paint.
• + Good skin texture on the man's hands and face.

• − Failed to include the requested motion blur from passing cars.
• − The hands are anatomically distorted with extra or fused fingers.
• − The framing feels less like a 'candid street photo' and more like a tight crop.

Seedream 4.5

• + Exceptional texture detail on skin, leather, and engraved metal.
• + Follows the 'close portrait' instruction perfectly with a tight, intimate crop.
• + Naturalistic lighting and shallow depth of field create a cinematic feel.

Vidu Q2

• + Excellent interpretation of the armored plate and leather straps.
• + Clean, sharp rendering of the hair braids and beads.
• + Good sense of environmental scale and character presence.

• − Composition is a medium shot rather than the requested 'close portrait'.
• − Lighting feels slightly flatter and more artificial compared to the prompt's warm torchlight request.
• − The skin textures appear smoother and less 'battle-worn' than Model A.

Seedream 4.5

• + Excellent photographic quality and realism in the food imagery.
• + Clean and legible layout with clear alignment between text and images.
• + Successfully followed the requested section headers.

• − Repeated the same category name as menu items (e.g., 'Pizza' listed as a pizza name).
• − Color-coded borders are a bit simplistic for 'modern minimalist' design.

Vidu Q2

• + Strong composition with a varied grid layout.
• + Offers a more comprehensive menu structure with descriptions and prices.
• + Creative use of geometric accents and vibrant design elements.

• − Text rendering is very poor with significant gibberish and spelling errors in the headers.
• − The food photos lack the depth and realism found in the competitor's image.
• − Frequent artifacts in the typography and logo areas.

Seedream 4.5

• + Excellent typography with a cohesive fire texture.
• + High-quality, photorealistic textures on the bun and patty.
• + Effective use of motion blur on peripheral ingredients.

• − The main burger stack is less 'exploded' and more assembled compared to the prompt.

Vidu Q2

• + Better 'exploded' composition with clear gaps between all layers.
• + Bright, vibrant colors that stand out against the fire background.
• + Includes dynamic sauce splashes.

• − The pricing text has a significant error, rendering a strange character instead of a Euro symbol.
• − The lighting on the burger components is a bit flat compared to the background.

Seedream 4.5

• + Excellent text legibility and spelling throughout.
• + Consistently clean and realistic chalk texture.
• + Well-organized layout with proper letter spacing.

• − Repeated the price '$24' on two lines for the first entry.
• − The cursive title is fairly standard rather than 'elegant cursive'.

Vidu Q2

• + Stronger visual sense of hand-drawn variation in letter heights.
• + More expressive 'chalk' strokes with heavy and light pressure.

• − Numerous spelling errors like 'Truffe Musshoom', 'Lemepun', and 'Octopd'.
• − Text becomes garbled and illegible at the bottom of the board.
• − The price for the first item is incorrect based on the prompt.

Seedream 4.5

• + Near-perfect adherence to the character reference, including the face, hair, and clothing.
• + Excellent integration of the scarf and accessories into the new pose.
• + Faithfully captures the lighting and environment of the source image.

• − The leg positioning is slightly simplified compared to the complex cross-legged pose in Image 1.
• − The toes on the left foot (viewer's left) are somewhat blurred.

Vidu Q2

• + Attempts the difficult leg-crossing pose from Image 1 more accurately than Model A.
• + High local detail on the face and clothing textures.

• − Significant anatomical errors, including a sixth finger on the right hand and an extra wrist/thumb joint on the left.
• − The legs appear detached or poorly blended with the torso.

Seedream 4.5

• + Excellent lighting and cinematic color grading.
• + The astronaut's pose and hand gesture feel dynamic.
• + Realistic horse anatomy and fur texture.

• − Completely failed the negative constraint to put the horse on top of the astronaut.
• − The composition is a standard trope rather than being surreal as requested.

Vidu Q2

• + Ornate, cosmic aesthetic for the horse's coat matches a surreal theme.
• + High level of detail in the space suit equipment.

• − Failed the specific prompt instruction to place the horse on top.
• − Unusual anatomical glitches where the horse's back leg meets the nebulae.

Seedream 4.5

• + Excellent preservation of the person's original identity, face, and vitiligo patterns
• + High fidelity in replicating the scarf, watch, ring, and coat details from Image 2
• + Superior integration of clothing onto the subject's pose and body shape

• − The black shirt under the coat has a strange cutout revealing the torso vitiligo that wasn't in the reference

Vidu Q2

• + Included the sunglasses and mustache from the source person in Image 2
• + Captured the texture of the fabric well

• − Failed to preserve the original person's identity, heavily altering his face and hair
• − Left vitiligo-like texture floating on the outside of the coat sleeve
• − Significantly changed the background environment

Seedream 4.5

• + Excellent text rendering on the taxi hat
• + Captured the 'bored' expression of the passenger perfectly
• + Stronger realistic fur texture on the capybara

• − The perspective makes the capybara look unusually large compared to the car interior
• − Lower visual clarity on the background city lights

Vidu Q2

• + Natural side-profile composition showing more of the car interior
• + Good lighting consistency across the passenger and driver
• + Dynamic and clear bokeh in the city background

• − The passenger's hands and phone interaction are slightly distorted
• − The capybara's hat is a generic chauffeur style rather than the requested yellow taxi cap
• − Capybara's paws look somewhat more like human/monkey hands than realistic rodent paws

Seedream 4.5

• + Excellent text rendering with perfect spelling and punctuation.
• + High-quality cinematic lighting around the jack-o-lantern.
• + Clean and professional layout that feels like a real digital invitation.

• − Missed the request for a dark parchment texture for the main poster background.
• − The web elements in the corners look a bit like generic clip-art.

Vidu Q2

• + Successfully incorporated the dark parchment poster texture.
• + Spooky border with webs and thorns matches the 'vintage' aesthetic well.
• + Good use of variety in the bat silhouettes.

• − Numerous spelling errors including 'Intoviztion', 'invieed', and 'fiigts'.
• − Date is incorrect (30.70.2025 instead of 30.10.2026).
• − Text is cramped and poorly aligned compared to the other model.

Seedream 4.5

• + Excellent preservation of the original facial features and lighting.
• + The hairline integration is seamless with the existing forehead and sideburns.
• + Perfect source preservation, maintaining identical background and clothing details.

• − The hair volume is a bit conservative given the 'thick head of hair' prompt.

Vidu Q2

• + Successfully added a very thick and voluminous head of hair as requested.
• + Realistic hair texture and color match for the existing beard.

• − Noticeable distortion of the forehead and eyebrow area compared to the source image.
• − Artifacts at the top of the hair where it meets the sky, looking a bit 'pasted on'.
• − Alters the eye shape and glasses frame slightly.

Seedream 4.5

• + Excellent 3D rendering with realistic PBR material effects on the salmon
• + Perfectly followed text instructions and layout
• + Clean, professional-looking isometric diorama with soft lighting

• − The 'JAPAN' text is slightly off-center compared to the 'SUSHI' text

Vidu Q2

• + Successfully rendered all requested text elements and the flag
• + Vibrant colors and high visual appeal for a cartoon style
• + Good variety of sushi types on the plate

• − The sushi models look slightly more artificial and less detailed than Image A
• − The lighting is a bit flat compared to the soft shadows in Image A

Seedream 4.5

• + Successfully creates a 'big head' caricature style while maintaining high-fidelity features from the source.
• + Perfectly balances all three requested elements: news desk/mic, hockey gear, and a dog.
• + Keeps the background environment remarkably consistent with the source image.

• − The transition from the 3D-style body to the realistic face texture is slightly jarring.
• − The dog at the desk is a bit stiff in its posing.

Vidu Q2

• + Adopts a classic comic-book caricature style that feels very energetic and humorous.
• + Creative inclusion of two different dogs, one wearing a hockey-style jersey.
• + Excellent integration of the hockey rink background while maintaining the news anchor theme.

• − The hands are poorly rendered, with a varying number of fingers and unnatural joints.
• − Loss of the specific facial identity from the source image in favor of a generic cartoon face.
• − Text on the microphone tag is slightly messy.

Seedream 4.5

• + Excellent depiction of lighting with distinct god rays and dew sparkles matching the prompt.
• + Dynamic composition that creates a sense of movement and 'tumbling' together.
• + High texture detail in the fur, particularly on the fox and kitten.

• − The fox's eyes appear slightly stylized and overly large compared to the more realistic puppy and kitten.
• − The rabbit is somewhat lost in the center background due to the shallow depth of field.

Vidu Q2

• + Successfully includes all requested animal types with clear visibility.
• + Bright and cheerful lighting that captures the 'wholesome vibe' well.
• + Sharp focus across the entire group of animals.

• − Includes two golden retrievers, which was not requested.
• − The animals appear somewhat static and placed rather than 'tumbling' together.
• − The butterfly wings lack realistic detail, appearing like simple stickers in some areas.

Seedream 4.5

• + Perfectly captures the Studio Ghibli art style with simple, iconic character designs.
• + Excellent source preservation, keeping the composition and character poses nearly identical to the original meme.
• + Successfully incorporates soft watercolor textures and gentle lighting that feel hand-painted.

• − The woman in red has a slightly generic facial expression compared to the original's blurred motion look.

Vidu Q2

• + Maintains a high level of detail in the clothing patterns and textures.
• + Accurately reproduces the specific facial expressions and features of the original people.
• + Good use of pastel colors and bright, airy lighting.

• − The style leans more toward generic modern webtoon/manhwa than the specific Ghibli aesthetic requested.
• − The character features are too realistic and don't adopt the simplified Ghibli character design language as well as Model A.

Seedream 4.5

• + Strong sense of dynamic motion through hair and depth-of-field blurring on leaves.
• + Successfully changes the pose of the woman to appear to be running/jogging.
• + Excellent integration of the dog into a faster-paced stride.

• − Changes the facial features significantly compared to the source image.
• − The background lighting is altered with a strong lens flare not present in the original.

Vidu Q2

• + Perfectly preserves the facial features and clothing of the subject from the source image.
• + High-quality, sharp rendering of the added leaves.
• + Successfully adds the requested hair motion while keeping the background stable.

• − The pose remains a walking pace, which clashes slightly with the intensity of the flying leaves.
• − Added leaves appear somewhat static and lack the motion blur seen in model A.

Seedream 4.5

• + Perfect text rendering for both the name and the date
• + Clean, balanced vector emblem composition
• + Excellent adherence to the requested color palette and vintage style

• − The steam icon is slightly simplified compared to the detail in the rest of the logo

Vidu Q2

• + Nice use of highlights and gradients on the cloche dome

• − Severe spelling errors in the primary name and date banner
• − Redundant and distorted text at the bottom of the image
• − Messy outlines around the banner and steam

Seedream 4.5

• + Excellent text rendering with no spelling errors.
• + Strict adherence to the requested timeline steps and iconography.
• + High-quality vector aesthetic with a professional layout.

• − Step 5 icon shows a satellite rather than a clear descending lunar module.
• − The Saturn V rocket proportions are slightly stylized/simplified.

Vidu Q2

• + Successfully captured the requested color palette.
• + Good use of the lunar surface in the composition.

• − Severe text garbling and frequent misspellings in every label.
• − Disconnected and confusing layout that fails to follow sequential logic.
• − Included 5 astronauts instead of the 3 involved in the mission.