Head to head: Recraft vs Ideogram V4.5 Edit
Recraft vs Ideogram V4.5 Edit
By Ryan Merket · Published
Eight image-editing challenges tested prompt fidelity across detailed scenes, constrained styles, spatial layouts, text, and exact object counts. The comparison turns on how consistently each model honored those specific instructions while still producing polished images.
Ideogram V4.5 Edit is the saved winner, with an aggregate score of **65.6** to Recraft’s **42.3**. It won seven tasks, Recraft won none, and exact counting was a tie. That makes this a **decisive result (not statistical)** on the saved task sample. The difference was clearest when prompts demanded several details at once. Ideogram more faithfully staged the futsal action and distinctive street signage, the dock sprint setup, and the centered goalie-mask display. It also followed the requested bedroom layout and flat-vector direction more closely, and rendered all three lines of poster text correctly. Recraft had appealing polish, especially in the rainy street scene and campsite, but those images drifted further from key instructions: the campsite departed from the restricted palette, and the futsal image missed distinctive prompt details. In exact counting, both models showed seven cups; Recraft’s presentation better matched the requested cups-only setup, while Ideogram added saucers. This assessment applies to the saved task sample, not to image models in general. **Verdict: Ideogram V4.5 Edit wins decisively on these eight prompts.**
Underpass futsal at neon rain
A cinematic street-sports scene, 16:9, painted in ultra-real digital realism: a late-night futsal match under the Kestrel Avenue overpass during fresh rain, lit by magenta ramen-shop signage, cyan pharmacy LEDs, and one harsh sodium-vapor streetlamp, with wet asphalt glowing in layered reflections; a player in a white jersey numbered 17 traps a yellow ball while steam rises from a storm drain, spectators with umbrellas cluster in deep shadow, and the mood is electric, humid, and tense. Strong contrast, motion frozen at the decisive instant, wide composition with dramatic neon night lighting.
OpenAI: GPT-6 Luna: Model B captures the decisive futsal action and more specifically includes the Kestrel Avenue sign, magenta ramen shop, cyan pharmacy lighting, sodium lamp, steam, and umbrella spectators. Model A has striking rain and reflections, but feels more like a street scene than a focused match and misses several of the prompt’s distinctive details. (Second judge pass, order swapped — scores are the average of both: Model B closely captures the requested decisive futsal moment, number 17, yellow ball, K | Anthropic: Claude Fable 5.1: Model B hits far more prompt specifics—Kestrel…
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 B more closely follows the flat-vector style and uses the requested palette, although its sky has a gradient and it includes two moons. Model A has a rich, balanced campsite scene, but its painterly texture, gradients, and many color variations conflict strongly with the restricted-palette brief. (Second judge pass, order swapped — scores are the average of both: Model B better respects the restricted four-color, flat-vector direction, though its sky uses a gradient and the campsite is qui | Anthropic: Claude Fable 5.1: Model A is a detailed, attractive scene but vio…
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 B follows the requested flat-vector style and places the bed, centered round rug, desk beneath the back-wall window, and floor lamp in the front-right corner. Model A has polished rendering, but it misses the floor lamp placement, puts the window on the left wall rather than the back wall, and is not flat-vector style. (Second judge pass, order swapped — scores are the average of both: Model B closely matches the requested flat-vector isometric layout, including the bed on the left, center | Anthropic: Claude Fable 5.1: Model B satisfies nearly every spatial constrai…
Legible multi-line text
A minimalist event poster with three lines of crisp, correctly-spelled text stacked and centered: 'NIGHT MARKET' large on top, 'Fridays · 6–11pm' in the middle, 'Riverside Pier 4' at the bottom, on a deep navy background, clean sans-serif, subtle grain.
OpenAI: GPT-6 Luna: Model B renders all three requested lines correctly in a clean, centered minimalist poster, though it adds a small “EST. 2024” footer. Model A has an appealing photographed sign, but the event-time line is incorrect and split into two lines, and the surrounding street scene departs from the poster brief. (Second judge pass, order swapped — scores are the average of both: Model B closely matches the minimalist poster prompt and renders the requested wording clearly, though it adds a small “EST. 2 | Anthropic: Claude Fable 5.1: Model B renders all three lines correctly ('Fri…
Climber gear macro detail
An extreme-close sports equipment portrait in high-end studio product photography: a chalk-dusted bouldering shoe named "Rift Talon X2" resting on a rough granite volume beside a coiled violet climbing rope and a brushed-metal belay device engraved "Aster Forge"; render the suede nap, rubber edge texture, woven laces, frayed stitching, powdered chalk specks, tiny granite crystals, and hairline scratches in the metal with immaculate crispness and no smudging. Controlled side lighting, dark neutral backdrop, shallow but precise depth of field focused on the shoe’s toe box.
OpenAI: GPT-6 Luna: Model B better matches the dark studio product portrait and includes the coiled violet rope and engraved metal belay device, with convincing suede, chalk, and granite detail. Model A is visually appealing, but its bright outdoor setting and missing belay device depart more from the prompt, and the “Aster Forge” text appears on the rope rather than on metal. (Second judge pass, order swapped — scores are the average of both: Model B better matches the controlled studio macro brief, with the grani | Anthropic: Claude Fable 5.1: Model A ignores the studio brief (outdoor dayli…
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 clearly separated, matching espresso cups in a straight overhead row on white marble, with no extra objects; Model B also has seven, but the added saucers and smaller cups depart somewhat from the requested presentation. Model A’s cups are more prominent and individually countable, while both entries have clean lighting and polished composition. (Second judge pass, order swapped — scores are the average of both: Model B shows exactly seven clearly separated, matching | Anthropic: Claude Fable 5.1: Both Model A and Model B render exactly seven es…
Dock sprint layout
A crisp editorial sports photograph, 16:9, of a dawn training drill on Pier 11 in Brinemouth: a female sprinter in a teal relay uniform is frozen mid-start on a red track lane laid over weathered wooden planks; exactly to her left sits a stack of three lime-green starting blocks, directly behind her stands a silver timing tripod with a small digital display, a bright orange duffel bag lies between the tripod and a white cooler, and a black carbon bicycle helmet rests on top of the cooler; beyond them, a coach in a yellow rain jacket stands behind the cooler pointing toward the water. Low camera angle, sharp realistic detail, cool sea mist, natural dawn light.
OpenAI: GPT-6 Luna: Model B includes nearly all the specified layout details, including the coach, cooler with helmet, digital timing tripod, and green starting blocks; Model A omits several of these and rearranges others. Model A is polished, but Model B is more faithful to the scene and still has a strong editorial composition. (Second judge pass, order swapped — scores are the average of both: Model B includes more of the requested dockside setup, including the timing tripod, orange duffel, cooler with helmet, y | Anthropic: Claude Fable 5.1: Model B includes nearly every requested element…
Goalie mask in trophy case
A hyper-real studio-style sports image, 16:9, of a scuffed cobalt-blue ice hockey goalie mask displayed inside a tall glass trophy case in a dim arena corridor; the front glass must show subtle reflections of red exit signs and overhead rink lights, the side panels reveal refracted views of a silver championship cup and a hanging maroon jersey outside the case, and the polished chrome cage of the mask mirrors the corridor with curved distortions; a faint reflection of a zamboni appears in the glossy floor beneath the case. Precise transparency, realistic internal reflections, cool arena ambience, dramatic centered composition.
OpenAI: GPT-6 Luna: Model B better captures the centered, cool-lit display with a visibly scuffed blue mask, hanging maroon jersey, trophy, and reflected exit sign. Model A has convincing corridor depth and reflections, but places the mask off-center and the cup inside the case; neither image clearly includes the requested zamboni reflection. (Second judge pass, order swapped — scores are the average of both: Model B more closely matches the centered, dim arena display, with a scuffed blue mask, red exit reflection | Anthropic: Claude Fable 5.1: Model B delivers the dramatic centered composit…
Matchup powered by OpenRouter.