Head to head: Seedream 5.0 Pro Image Editing vs Hidream I1 Full
Seedream 5.0 Pro Image Editing vs Hidream I1 Full
By Ryan Merket · Published
Eight image tasks put these models through exact counting, spatial placement, material realism, and tightly constrained art direction. The results reveal a sharp difference in how reliably each follows detailed prompts.
Seedream won all eight tasks, with no ties, and its 66.7 aggregate score beat Hidream I1 Full’s 43.5. The verdict is decisive at limited confidence—not because every Seedream image was flawless, but because it repeatedly preserved the details Hidream dropped or blurred. The clearest gaps came on literal requirements. Seedream showed seven cups where Hidream showed five; its bedroom layout included the specified floor lamp and desk placement, while Hidream missed both. In the atrium scene, Seedream carried through the suspended sneaker, phone-case kiosk, café, yellow-column ad, and shoppers. Hidream supplied a convincing mall and a striking yellow bag, but left out several of those requested details. Seedream also came out ahead on visual specificity: its rainy storefront made the Velvet Current sign legible and reflections convincing, while its chrome-and-water still life showed the apple and window reflections more clearly. The macro lip-balm image read as exposed balm with a droplet; Hidream’s looked more like a closed tube. For the restricted-palette campsite, Seedream stayed closer to the four-color flat-vector brief, while Hidream leaned on gradients and extra shades. Neither model nailed every instruction—Seedream’s breakfast croissant was whole, and its basil pot was partly cropped—but those were isolated misses against otherwise faithful compositions. **Verdict: Seedream 5.0 Pro Image Editing wins decisively, 8–0, by following the brief more completely across both precise constraints and complex scenes.**
Exact counting
A clean overhead flat lay on white marble of EXACTLY seven identical espresso cups arranged in a single straight row, each clearly separate and individually countable, even soft lighting, 16:9.
OpenAI: GPT-6 Luna: Model A shows exactly seven individually countable cups in a straight row on white marble, closely matching the prompt. Model B is also a clean overhead marble flat lay, but it shows only five cups rather than seven. (Second judge pass, order swapped — scores are the average of both: Model B shows only five cups, missing the exact-count requirement, while Model A clearly presents seven separate, identical cups in a straight row on white marble. Model A also has clean overhead framing and even li | Anthropic: Claude Fable 5.1: Model A shows exactly seven identical espresso…
Atrium escalator scale test
A realistic interior of a six-story outlet atrium with a skylight, seen from the third-floor railing so the perspective lines converge cleanly toward the far café; in the foreground a giant yellow shopping bag advertisement wraps a column, midground shoppers carry ordinary consumer products like boxed blenders and rolled rugs, and in the distance a child-sized kiosk sells phone cases beneath a much larger suspended sneaker display, all objects sized convincingly relative to one another; precise architectural perspective, no warped geometry, natural midday light, contemporary commercial visualization, 16:9.
OpenAI: GPT-6 Luna: Model A includes the skylit multi-story atrium, a yellow column advertisement, a conspicuous suspended sneaker, a distant café, a phone-case kiosk, and shoppers carrying recognizable consumer goods, with strong converging perspective. Model B has a striking yellow bag and mall setting, but omits several key requested details and its merchandise displays do not clearly match the phone-case kiosk or suspended sneaker display. (Second judge pass, order swapped — scores are the average of both: Mode | Anthropic: Claude Fable 5.1: Model A delivers nearly every prompt element —…
Rainy storefront realism
A photorealistic street-level scene of a small headphone repair shop called "Velvet Current" at dusk after rain, viewed through the open doorway: brushed aluminum over-ear headphones on a walnut workbench beside a folded charcoal hoodie, a technician’s hand with natural skin texture adjusting a copper screwdriver near a matte-black amplifier, wet pavement outside reflecting the neon sign, believable mixed lighting from cool blue rainlight and a warm tungsten desk lamp, physically accurate shadows, reflections, fabric weave, wood grain, metal sheen, and skin translucency, cinematic realism, 16:9.
OpenAI: GPT-6 Luna: Model A more completely captures the prompt, with clearly legible “Velvet Current” signage, convincing rainy reflections, and a strong warm/cool lighting contrast around the workbench. Model B includes the key objects and rainy storefront, but its sign is less legible and the close-up hand and tools look less natural. (Second judge pass, order swapped — scores are the average of both: Model A more completely matches the requested scene, with clearly rendered “Velvet Current” signage, visible rai | Anthropic: Claude Fable 5.1: Model A delivers nearly every element—clean 'Ve…
Reflections & glass
A photorealistic close-up of a chrome teapot on a polished dark marble counter next to a glass of water; the window and a red apple on the counter must be correctly reflected in BOTH the chrome and the water's surface, dramatic side light, 16:9.
OpenAI: GPT-6 Luna: Model A delivers the requested close-up and dramatic side light, with the window and red apple visibly represented in the chrome and glass water. Model B is also polished and includes the key objects, but the reflections in the water are less distinct and the overall framing feels less close-up. (Second judge pass, order swapped — scores are the average of both: Model A presents a stronger close-up with clear window and apple reflections in the chrome, a visible apple reflection through the wate | Anthropic: Claude Fable 5.1: Model A clearly shows the window and red apple…
Macro lip balm bead
An extreme macro close-up of a twisted-up pearly peach lip balm in a translucent frosted tube lying on scattered sugar crystals, with one tiny condensed water droplet clinging to the balm edge and a faint fingerprint on the plastic cap beside it; beauty-product advertising photography, ultra-shallow depth of field, creamy circular bokeh, 100mm macro lens look, razor-thin focus on the droplet and wax texture, soft diffused studio light with a blush-pink backdrop.
OpenAI: GPT-6 Luna: Model A more directly depicts the exposed pearly peach balm with a droplet at its edge, frosted packaging, sugar crystals, and blush-pink bokeh, though the tube geometry is somewhat ambiguous. Model B is polished and appealing but reads more like a closed balm tube and misses the exposed twisted-up balm and fingerprint detail. (Second judge pass, order swapped — scores are the average of both: Model A more clearly shows the exposed, pearly peach balm in a frosted tube, with the droplet, sugar cr | Anthropic: Claude Fable 5.1: Model A delivers a convincingly twisted-up pear…
Spatial layout
A clean isometric illustration of a bedroom: a bed against the LEFT wall, a round rug centered on the floor, a desk under the WINDOW on the back wall, and a floor lamp in the FRONT-RIGHT corner. Flat-vector style, consistent perspective.
OpenAI: GPT-6 Luna: Model A places the bed, centered round rug, desk beneath the back-wall window, and floor lamp in the front-right corner as requested, with a polished flat-vector look. Model B has a clear isometric layout, but its front-right corner lacks a floor lamp and the desk is not clearly positioned beneath the window. (Second judge pass, order swapped — scores are the average of both: Model B has the cleaner isometric perspective, but it lacks the requested floor lamp in the front-right corner and places | Anthropic: Claude Fable 5.1: Model A places every element as specified — bed…
Studio breakfast assembly
A bright editorial product scene showing seven everyday breakfast items interacting in one coherent moment on a mint-green kitchen island: a coral electric kettle pouring into a glass mug, a chrome toaster ejecting two uneven slices, an open apricot jam jar with the lid tilted against it, a butter knife resting across a ceramic plate with half a croissant, a striped dish towel partly tucked under a wooden cutting board, and a tiny potted basil plant leaning toward the steam; photorealistic commercial style, balanced triangular composition with clear spacing and believable overlaps, soft morning window light from the left, crisp detail, 16:9.
OpenAI: GPT-6 Luna: Model A presents a polished, coherent breakfast scene with nearly all requested items, though the croissant is whole rather than half and the basil pot is mostly out of frame. Model B includes more extraneous food and duplicate jars, and its kettle-to-mug interaction is less convincing, making the composition feel less faithful and cohesive. (Second judge pass, order swapped — scores are the average of both: Model A includes nearly all requested objects and interactions, with the towel tucked un | Anthropic: Claude Fable 5.1: Model A includes nearly every requested element…
Restricted palette
A flat-vector illustration of a mountain campsite at dusk using ONLY four colors — cream, burnt orange, teal, and charcoal — with no gradients and no stray off-palette colors, balanced composition, 16:9.
OpenAI: GPT-6 Luna: Model A better matches the flat-vector style and tightly restricted four-color palette while clearly depicting a campsite at dusk; Model B has appealing depth but relies on gradients and numerous off-palette shades. Model A’s scene is detailed and balanced overall, though its foreground is somewhat crowded. (Second judge pass, order swapped — scores are the average of both: Model B has a strong, balanced campsite scene, but its gradients and many color variations conflict with the strict four-co | Anthropic: Claude Fable 5.1: Model A holds strictly to the four-color flat-v…
Matchup powered by OpenRouter.