Head to head: Kling Image vs Ideogram V4.5 Edit
Kling Image vs Ideogram V4.5 Edit
By Ryan Merket · Published
A seven-task image-model comparison puts cinematic action, exact-count composition, and visual style under the same microscope. The results turn on prompt priorities: framing and realism often pull one way, polish and stylistic interpretation another.
This is a tie, not a narrow win dressed up as a verdict. Kling Image scores 55.5 to Ideogram V4.5 Edit’s 59.9, but the judges put the odds that either is genuinely better at just 64%—too close to call. The small aggregate edge doesn’t survive the task-by-task split: Kling takes three, Ideogram takes three, and exact counting is even. Kling’s strongest work is in scenes that need motion, depth, or a specific wide composition. Its courier drone leap has the storm, rotor-wash spray, and handoff energy Ideogram’s calmer version lacks; it also handles the warm, one-point-perspective aisle and the cable-perched owl scene more faithfully to their requested framing and details. Ideogram is more persuasive when the brief hinges on a strong visual language or a polished arrangement. It better evokes carved ukiyo-e linework in the rainy server alley and the cresting-wave image, and its bound-materials still life keeps the objects clearer. But those wins come with recurring compromises: square framing where 16:9 was requested, or a missed detail such as the figure-eight cable coil. Neither model is reliably better at following every constraint. In the seven-cup test, Ideogram makes the count clearer while Kling preserves the widescreen format but crops the end cups; in the restricted-palette campsite, the judges split over Ideogram’s cohesive colors versus Kling’s closer framing and flatter shapes. **Final call: TOO CLOSE TO CALL. These models are effectively even here; choose by task, not by the aggregate score.**
Ukiyo-e server rain
A nighttime alley lined with humming micro-server kiosks and hanging fiber bundles rendered faithfully in traditional ukiyo-e woodblock style, with indigo rain slanting across the scene, a repair woman in a conical straw hat checking a glowing handheld network meter, flattened perspective, carved linework, visible woodgrain texture, limited Edo-era pigment palette translated to futuristic technology, strong diagonal composition, lantern-like highlights reflecting in puddles, 16:9.
OpenAI: GPT-6 Luna: Model B more convincingly captures traditional ukiyo-e through flattened perspective, bold carved linework, a limited indigo-and-coral palette, and prominent hanging fiber bundles. Model A depicts the futuristic kiosks and rainy alley clearly and matches the requested widescreen format, but its more realistic rendering is less faithful to the woodblock style. (Second judge pass, order swapped — scores are the average of both: Model B better captures the carved-line ukiyo-e look, indigo-and-coral | Anthropic: Claude Fable 5.1: Model A delivers the requested 16:9 frame, a ri…
Courier drone leap
A compact rescue courier drone with six translucent rotors bursts upward from the rain-slick deck of an offshore data-buoy station, mid-action as it snatches a yellow emergency SSD capsule from a gloved technician’s outstretched hand; rotor wash sprays crescent sheets of seawater, loose cable ties whip sideways, one landing leg is still skimming the deck, and the drone’s nose light cuts through mist with convincing speed and tension, cinematic photojournalism style, low angle, stormy dawn backlight, sharp frozen droplets, 16:9.
OpenAI: GPT-6 Luna: Model A better captures the stormy, low-angle burst of motion, with rain, dramatic rotor-wash spray, a gloved hand presenting the yellow capsule, and the requested widescreen framing. Model B clearly shows six rotors and a yellow payload, but feels calmer and more posed, and its handoff is less convincing. (Second judge pass, order swapped — scores are the average of both: Model A better captures the stormy, low-angle cinematic action with dramatic rotor-wash spray, rain, and a convincing techni | Anthropic: Claude Fable 5.1: Model A delivers the requested 16:9 cinematic l…
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 seven cups in a 16:9 row, but the end cups are cropped and less clearly separate. Model B makes all seven cups easy to count in a clean overhead arrangement, though it misses the requested 16:9 format. (Second judge pass, order swapped — scores are the average of both: Model B shows exactly seven complete, clearly separated identical cups in a straight row on white marble. Model A uses the requested widescreen framing, but cups are cropped at both edges and the row is less cleanly | Anthropic: Claude Fable 5.1: Both Model A and Model B show exactly seven espr…
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 matches the requested 16:9 flat-vector campsite scene more closely, though its lantern and tent introduce extra shades beyond the strict palette. Model B is more visually elaborate, but its square format and broader range of color shades reduce prompt adherence. (Second judge pass, order swapped — scores are the average of both: Model B better captures a dusk campsite with a cohesive restricted-palette look and a richly balanced scene, though it misses the requested 16:9 format. Model A | Anthropic: Claude Fable 5.1: Model A delivers the 16:9 frame and flat shapes…
Perspective & scale
A photorealistic one-point-perspective shot down a long empty library aisle, shelves receding to a single vanishing point, books getting consistently smaller with distance and nothing warped or wrongly sized, warm overhead light, 16:9.
OpenAI: GPT-6 Luna: Model A more fully matches the requested warm, photorealistic 16:9 aisle, with clear one-point perspective and convincing scale recession. Model B has a strong centered vanishing point, but its square framing and cooler lighting depart from the prompt. (Second judge pass, order swapped — scores are the average of both: Model A better matches the requested 16:9 format and warm overhead lighting while maintaining a convincing, centered one-point perspective. Model B also has strong symmetry and re | Anthropic: Claude Fable 5.1: Model A delivers a true 16:9 photorealistic one…
Named art style
A ukiyo-e woodblock print of a fishing boat riding a large cresting wave at dawn, faithful to the flat color planes, bold outlines, and stylized foam of the tradition, muted indigo and cream palette.
OpenAI: GPT-6 Luna: Model B more convincingly evokes ukiyo-e through its dramatic cresting wave, stylized foam, and strong woodblock-like outlines, while clearly suggesting dawn with the rising sun. Model A better sticks to the muted indigo-and-cream palette, but Model B’s composition is more dynamic and polished despite using some brighter colors than requested. (Second judge pass, order swapped — scores are the average of both: Model B more fully captures the dawn setting, dramatic cresting wave, and bold ukiyo-e | Anthropic: Claude Fable 5.1: Model B delivers a true large cresting wave wit…
Bound materials still life
A precise studio still life of five prototype desk gadgets arranged on matte charcoal paper: a cobalt-blue anodized aluminum voice recorder, a saffron-yellow silicone earbud case, a jade-green frosted glass smart candle, a crimson braided nylon charging cable coiled in a figure-eight, and a white ceramic palm-sized Wi-Fi sensor; each object must keep its own exact color and material with no swapping or bleeding, ultra-clean product photography style, top-down three-quarter composition, soft polarised lighting with crisp shadow edges, 16:9.
OpenAI: GPT-6 Luna: Model B shows all five objects clearly in a polished studio arrangement, while Model A crops the candle and cable and gives the white sensor an unusual hand-like shape. Model B also misses the figure-eight cable coil and 16:9 framing, but follows the prompt more completely overall. (Second judge pass, order swapped — scores are the average of both: Model A better matches the requested 16:9 studio arrangement and gives all five objects clear space, while Model B is polished but square and its cab | Anthropic: Claude Fable 5.1: Model A hits the 16:9 ratio but crops the candl…
Melancholy transit map owl
A clean stylized vector illustration of a small maintenance robot shaped like an owl perched on a suspended tram cable above a quiet tech district, holding a tiny inspection tablet, mood of midnight solitude and gentle vigilance; simplified geometric forms, smooth curves, minimal linework, controlled teal-plum-amber palette, subtle fog layers, ample negative space, editorial illustration feel with crisp edges and a balanced off-center composition, 16:9.
OpenAI: GPT-6 Luna: Model A more clearly shows an owl-shaped maintenance robot perched on a cable above a tech district, with a tablet and the requested wide editorial composition. Model B has a quieter, more solitary feel, but its square framing and extra background robot weaken the prompt fit. (Second judge pass, order swapped — scores are the average of both: Model A better matches the requested 16:9 editorial composition, with a cable-perched owl, inspection tablet, layered tech district, and a restrained plum- | Anthropic: Claude Fable 5.1: Model A delivers the requested 16:9 format, a b…
Matchup powered by OpenRouter.