HiDream I1 Fast HiDream AI vs Qwen Image Alibaba

HiDream I1 Fast

• + Excellent handling of glass transparency and reflections
• + Very high textural detail on the book and table
• + Realistic interaction of light through the glass

• − The sphere is quite large relative to the 'small' descriptor
• − The red book appears more like a flat box or thick cover without visible pages

Qwen Image

• + Perfect adherence to the 'small' blue sphere instruction
• + Highly realistic book rendering with visible pages
• + Strong overall composition and depth of field

• − The glass cube has internal vertical lines that look like supports rather than a single glass volume
• − The plant is less visible through the glass compared to Model A

HiDream I1 Fast

• + Excellent shallow depth of field and bokeh
• + Subtle, realistic motion blur in the background car
• + Detailed reflections on the wet pavement

• − Anatomical issue with the man appearing to sit on the bicycle frame in an unnatural position
• − The bike geometry is slightly warped near the pedals

Qwen Image

• + Natural standing posture and gesture of repairing/adjusting the bike
• + Clearly visible raindrops as requested
• + Composition feels more grounded and realistic to a street photography scene

• − The white car in the background lacks the requested motion blur
• − Less cinematic lighting compared to Model A

HiDream I1 Fast

• + Excellent engraving details on the plate armor.
• + Warm torchlight lighting is beautifully integrated with the skin tones.
• + Portrait composition is strong with a centered, heroic focus.

• − The scars look a bit like face paint rather than physical wounds.
• − The hair braids appear slightly stiff and unnatural in their symmetrical placement.

Qwen Image

• + Highly realistic representation of multiple leather straps and chainmail textures.
• + The sparks and torchlight effect are more literal and dynamic.
• + The 'battle-worn' aspect is more convincing with varied, textured scars and blood splatter.

• − The sparks have a somewhat digital, artificial look compared to organic bokeh.
• − The armor engraving is slightly less detailed than the competing model.

HiDream I1 Fast

• + Strong category visibility with a bold red accent bar.
• + Consistent food photography style across the grid.
• + High-contrast header that stands out.

• − Poor typography rendering with significant artifacts and illegible characters.
• − Layout feels cramped with limited white space.
• − The main title is garbled and distracting.

Qwen Image

• + Clean, modern aesthetic with excellent use of white space.
• + Better font clarity and professional layout structure.
• + Photography is vibrant and fits the 'casual dining' brief perfectly.

• − Text is still mostly nonsensical/placeholder despite better character forms.
• − The grid layout of images feels slightly repetitive.

HiDream I1 Fast

• + Excellent font rendering for the main title with a high-quality glow effect.
• + Accurate rendering of the requested price in a starburst.
• + Clean, professional-looking food photography style.

• − Failed the 'exploded' instruction as the burger is mostly assembled.
• − Includes redundant, garbled text in the secondary price circle.

Qwen Image

• + Successfully captured the 'exploded' motion with flying ingredients.
• + Stronger dynamic composition with varying depths of field.
• + Consistent fiery glow applied across all requested text elements.

• − The text 'LIMITED TIME ONLY' is slightly thinner and less impactful than Model A.
• − A few small artifacts in the flying sauce droplets.

HiDream I1 Fast

• + Natural chalk texture on the 'Today's Specials' title
• + Followed the correct year (2026)

• − Severe text artifacts and overlapping words in the menu items
• − Spelling errors like 'Herbs' joined with numbers and 'Buter'
• − Messy, disorganized layout with random price tags floating

Qwen Image

• + Clear, legible handwriting across the entire board
• + Excellent layout and composition that fills the chalkboard well
• + Correct completion of the truncated prompt 'Brown But...' into 'Brown Butter Chocolate Chip Cookies'

• − Incorrect year (20026 instead of 2026)
• − The 'handwriting' looks slightly more like a digital font than natural chalk

HiDream I1 Fast

• + High resolution and realistic textures on the horse's coat and sand.
• + Natural lighting and cinematic color grading.

• − Completely failed the environment prompt, placing the subjects in a desert instead of space.
• − Failed the counter-intuitive instruction 'horse on top, not vice versa'.

Qwen Image

• + Successfully placed the subjects in a space environment with planets and Earth visible.
• + Clean anatomical rendering of both the astronaut and the horse.

• − Failed the counter-intuitive instruction 'horse on top, not vice versa'.
• − The floating horse lacks specific surreal elements to make it 'rideable' in a vacuum.

HiDream I1 Fast

• + Excellent texture on the capybara's fur
• + Good 'taxi' branding on the cap
• + Cinematic lighting consistent with a night scene

• − The driver has human hands instead of capybara paws
• − The passenger is sitting in the front seat instead of the back seat
• − Anatomy of the capybara head merged with a human torso is jarring

Qwen Image

• + Correctly places the businesswoman in the back seat
• + Successfully renders animal paws on the steering wheel
• + Captures the 'bored' expression of the passenger perfectly

• − The text on the taxi sign says 'YOXI' instead of 'TAXI'
• − Minor lighting inconsistencies on the passenger's face

HiDream I1 Fast

• + Successfully integrated large, readable gothic headers
• + Captures a fun, vintage illustration style
• + Correctly included the central glowing jack-o-lantern

• − Text in the scroll and the bottom details contains typos and garbled characters
• − Border lacks the requested thorns
• − Overall composition feels a bit crowded with overlapping elements

Qwen Image

• + Excellent adherence to the 'thorns and webs' border prompt
• + Superior text rendering for the event details at the bottom
• + Atmospheric cinematic lighting and polished composition

• − Main gothic title text has overlapping and misaligned characters
• − The text 'to a night of frights' is slightly floating above the scroll rather than on it

HiDream I1 Fast

• + Excellent text rendering with clean, professional typography.
• + Very high-fidelity textures, specifically on the rice and salmon textures.
• + Accurate adherence to the square layout and isometric 45° angle.

• − The sushi roll inside the nigiri piece is a structural inaccuracy.
• − Minor artifacts near the sushi graphic adjacent to the text.

Qwen Image

• + Successfully includes the requested Japan flag icon in the text area and as a physical prop.
• + Charming miniature aesthetic that hits the cartoon/diorama brief perfectly.
• + Clean, balanced composition with soft, appealing lighting.

• − Text is slightly less crisp and professional than Model A.
• − The 'N' in 'JAPAN' has a subtle typographical artifact.
• − Chopsticks appear slightly merged with the plate.

HiDream I1 Fast

• + Excellent lighting and warm golden hour atmosphere.
• + Highly detailed fur textures on the animals.

• − Failed to include the requested baby bunny.
• − Included two kittens instead of one, missing a variety requirement.
• − The animals are sitting still rather than 'playfully chasing' or 'tumbling'.

Qwen Image

• + Successfully included all four requested animals: puppy, kitten, bunny, and fox kit.
• + Better captured the motion of chasing and playing as requested in the prompt.
• + Superior rendering of 'god rays' and dew sparkles.

• − The fox kit has slightly anatomically unusual front paws.
• − The kitten's tail is a bit thick and blurry compared to the rest of the body.

HiDream I1 Fast

• + Clean vector emblem layout
• + Good use of negative space in the cloche

• − Typos in text including extra digits
• − Inconsistent steam lines

Qwen Image

• + Correct date on banner
• + Consistent vintage texture and coloring

• − Messy overlapping letter forms
• − Unbalanced central composition

HiDream I1 Fast

• + Strong adherence to the requested NASA-inspired color palette.
• + Good central composition with the Saturn V rocket.

• − Text rendering is very poor with significant gibberish and misspellings.
• − Iconography does not logically match the labels (e.g., Saturn icon for Translunar).
• − The sequence of steps is disorganized and difficult to follow.

Qwen Image

• + Excellent vector art style that is clean and modern.
• + Follows the structural flow of an infographic much better than Model A.
• + Includes better iconography for the Lunar Module and mission crew.

• − Text contains several spelling errors (e.g., 'Sarth', 'ApolL', 'Aldlin').
• − Failed to include all 6 specifically requested steps in numerical order.