Head to head: Kling Image vs Ideogram V4.5 Edit
Kling Image vs Ideogram V4.5 Edit
By Ryan Merket · Published
This matchup tests Kling Image and Ideogram V4.5 Edit across reflective scenes, strict palettes, perspective, product photography, and exact-count prompts. The results turn on different kinds of prompt fidelity: some tasks reward a tightly controlled composition, others strict adherence to visual constraints.
Ideogram V4.5 Edit is the saved winner, with an aggregate score of **60.5** to Kling Image’s **56.0** and a **slight edge (not statistical)**. The task record is close in shape: Ideogram leads on four tasks, Kling on two, with two ties. Kling’s strongest showings are the reflections-and-glass close-up and the empty bookstore aisle. It makes the teapot, glass, and apple more prominent amid reflected light, and its aisle has the cleaner one-point perspective. Ideogram, by contrast, is more reliable when the brief imposes tight constraints: it handles the restricted palettes better, presents the seaside kiosk more completely, and makes all seven cups easy to count. The ties show where neither model fully owns the brief. For the dumplings, Ideogram is closer to the requested count, while both images satisfy enough of the scene requirements to split the task verdict. On the yuzu fizz, Kling gives condensation and fizz more emphasis; Ideogram’s restrained product styling and readable label pull the other way. **Final call: Ideogram V4.5 Edit takes this matchup by a slight edge, not a runaway.** Its advantage comes from more consistent constraint-following and clearer completeness on several prompts, while Kling has two distinct wins in composition and perspective. This assessment applies to the saved task sample.
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 more clearly delivers the close-up arrangement and shows the window and apple reflected in the chrome and water, with striking dramatic side light. Model B has convincing chrome and window reflections, but the subjects are smaller in the frame and the required reflections in the water are less evident. (Second judge pass, order swapped — scores are the average of both: Model A makes the teapot, glass, and apple more prominent and shows clearer window and apple reflections on the chrome, | Anthropic: Claude Fable 5.1: Model A delivers a genuinely photorealistic scen…
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 adheres more closely to the four-color restriction and presents a clean, flat-vector campsite with clear mountain and dusk elements. Model A has a strong campsite composition and crescent moon, but its tent and ground introduce several colors beyond the specified palette. (Second judge pass, order swapped — scores are the average of both: Model B follows the four-color restriction closely and presents a clear, balanced flat-vector campsite at dusk. Model A has a strong central scene, but | Anthropic: Claude Fable 5.1: Model A breaks the four-color constraint with b…
Count the dumplings
A candid editorial food photograph of a midnight street-food stall, showing EXACTLY 17 distinct saffron-yellow soup dumplings arranged across three bamboo steamers on a brushed steel counter, every dumpling fully visible and individually countable with none overlapping or hidden; include a chipped ceramic saucer of black vinegar, a handwritten price tag reading 'Lot 43', and drifting steam under cool fluorescent night-market lighting, shallow depth only in the background, hyper-realistic, 16:9.
OpenAI: GPT-6 Luna: Model B appears closer to the requested count of 17, while Model A visibly includes substantially more dumplings; both show three steamers, steam, vinegar, and the requested tag. Model A has a clearer tag and balanced layout, but Model B better matches the saffron-yellow color and atmospheric night-market feel. (Second judge pass, order swapped — scores are the average of both: Model B is closer to the requested dumpling count and captures the moody night-market setting, steam, and yellow dumpli | Anthropic: Claude Fable 5.1: Model B lands much closer to the required 17 du…
Glass kiosk reflections
A hyper-realistic dawn scene of a tiny seaside espresso kiosk built from glass panels and polished chrome, viewed from outside at a slight angle so the front window, side window, chrome espresso machine, and a rain puddle all simultaneously reflect and refract the surroundings correctly: a striped awning, stacked blood-orange pastries, the barista’s hands, gulls crossing the pink sky, and moored teal fishing boats warped appropriately by each surface; crisp natural light after rain, meticulous optical realism, cinematic composition, 16:9.
OpenAI: GPT-6 Luna: Model B presents the full seaside kiosk and more clearly includes the chrome machine, pastries, barista’s hands, gulls, teal boats, and puddle reflections in a cohesive cinematic scene. Model A has lovely dawn light and strong reflections, but the kiosk and espresso setup are more obscured by the tight framing. (Second judge pass, order swapped — scores are the average of both: Model B more completely delivers the requested angled view of both kiosk windows, chrome machine, pastries, barista’s h | Anthropic: Claude Fable 5.1: Model B delivers a cinematic, photographic dawn…
Aegean pastry palette
A stylized gourmet still life of braided sesame bread, fig jam in a shallow bowl, and two small cups of herbal tea arranged on a stone table in a modern Mediterranean illustration style, lit by clean noon sunlight with crisp shadows, restricted STRICTLY to the named color palette 'Aegean Dusk': #16324F deep navy, #2E5E6E muted teal, #C97B63 terracotta rose, #E7D8C9 shell beige, and #F2EDE3 soft chalk, with no other colors anywhere in the image, 16:9.
OpenAI: GPT-6 Luna: Model B follows the required Aegean palette much more closely and has a clean, cohesive Mediterranean illustration style, though its bowl shows whole figs rather than jam. Model A depicts the requested jam and bread more literally, but its many out-of-palette colors substantially conflict with the strict palette requirement. (Second judge pass, order swapped — scores are the average of both: Model B follows the restricted Aegean palette and modern illustration style much more closely, though the | Anthropic: Claude Fable 5.1: Model A ignores the strict 'Aegean Dusk' palett…
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 precisely delivers the empty, centered one-point-perspective aisle, with shelves and books receding consistently under warm overhead lighting. Model B is atmospheric, but the book on the floor and ladder detract from the requested empty aisle. (Second judge pass, order swapped — scores are the average of both: Model A has a clearer, more consistent one-point perspective, with shelves and overhead lights receding cleanly toward a single vanishing point and books scaling naturally wit | Anthropic: Claude Fable 5.1: Model A delivers a clean, empty one-point-persp…
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 B shows exactly seven identical cups, fully visible and individually countable in a straight row. Model A also appears to have seven, but the cups at both edges are cropped, making the arrangement less clear; Model B’s extra spoons are a minor deviation. (Second judge pass, order swapped — scores are the average of both: Model B shows exactly seven fully visible, clearly separated cups in a straight row, while Model A crops the cups at both edges, making the exact count less unambiguous. M | Anthropic: Claude Fable 5.1: Model A shows five full cups plus two cropped p…
Macro yuzu fizz
A macro close-up product photograph of a single chilled yuzu soda in a ribbed glass bottle just uncapped, shot at table level with a true macro lens effect: razor-sharp condensation beads and a tiny crown of fizz at the mouth of the bottle, while the background falls into creamy blur showing only vague hints of a lacquer tray, one sliced kumquat, and a linen napkin; dramatic side lighting, elegant Japanese café styling, ultra-detailed commercial realism.
OpenAI: GPT-6 Luna: Model A more strongly delivers the macro emphasis, sharply visible condensation, and fizz at the uncapped mouth, with the requested café props softened in the background. Model B has polished Japanese product styling and a readable label, but the condensation and fizz are less prominent and the framing feels less macro. (Second judge pass, order swapped — scores are the average of both: Model B more closely matches the restrained macro product-photo brief, with the uncapped ribbed bottle, fine c | Anthropic: Claude Fable 5.1: Model A delivers the requested true macro feel…
Matchup powered by OpenRouter.