Photorealism in Hunyuan Image 3.0 comes down to writing like a photographer, not a wish list. Tencent's open-source model — a native multimodal, unified autoregressive model with 80B parameters (about 13B activated) across a 64-expert mixture — has strong photoreal training and deep world knowledge, so it renders convincing photos when you hand it real photographic language: a named lens, an aperture, a specific light setup, and honest color and texture cues.
This guide teaches the pattern first, then gives you 10 complete copy-paste prompts to adapt. Use the Hunyuan Image cheat sheet for the full modifier tables, grab the ready-made Hunyuan photorealism pack for more scenes, and see the best Hunyuan Image prompts roundup for every other use case. One rule up front: this is a 2K model, so never ask for 4K.
The photorealism formula
Every photoreal prompt below follows the same six-part pattern, written as one flowing sentence in this order:
[Subject + appearance] + [Action/pose] + [Environment/setting] + [Lighting] + [Camera & lens] + [Style/quality cues]
Here is what goes in each slot and the concrete words that work:
| Part | What it does | Option words |
|---|---|---|
| Subject + appearance | Names who or what, with specific detail | age, clothing, material, color, condition (e.g. weathered, frosted, matte) |
| Action / pose | Gives a moment instead of a mannequin | mid-motion, looking off-camera, three-quarter view, resting on |
| Environment / setting | Places the subject in a real space | studio, sunlit kitchen, rainy street, coastal cliff, dark slate surface |
| Lighting | The single biggest realism lever | golden hour, soft window light, three-point, chiaroscuro, high-key, low-key, overcast diffusion |
| Camera & lens | Encodes a learned photographic look | 24mm, 35mm, 50mm, 85mm, 100mm macro; f/1.4, f/1.8, f/2.8, f/8 |
| Style / quality cues | Locks the medium and finish | photorealistic, Kodak Portra tones, subtle grain, shallow DOF, fine skin texture, true color |
You do not need every slot every time, but subject, lighting, and lens are the load-bearing three. Fill those and Hunyuan's world knowledge handles most of the rest.
Camera & lens language
A focal length and aperture are the fastest way to steer the look, because each maps to real photographs the model has seen. A 24mm wide-angle gives sweeping landscapes and interiors with deep space; a 35mm is the natural documentary and environmental-portrait lens; a 50mm reads like the human eye; an 85mm compresses and flatters faces for portraits; and a 100mm macro isolates fine texture on products and close-ups.
Aperture controls depth of field. f/1.4 to f/1.8 gives a shallow plane of focus and soft, creamy background blur — the candid and portrait look. f/2.8 is a balanced middle. f/8 keeps a scene sharp front to back, which is what you want for landscapes and architecture. Pairing a lens with an aperture — "85mm at f1.8" — is a compact, unambiguous instruction that does more work than a paragraph of adjectives.
Lighting that reads real
Vague light is the most common reason an image looks fake. Name a real setup and a direction instead. The workhorses:
- Golden hour — warm, low, raking sunlight with long shadows; the easiest instant-realism cue.
- Soft window light — directional but gentle; flattering for portraits, food, and interiors.
- Three-point lighting — clean studio standard with key, fill, and rim; reads as professional.
- Chiaroscuro / low-key — deep shadows and one bright highlight; dramatic and cinematic.
- High-key — bright, even, near-shadowless; beauty and clean commercial work.
- Overcast diffusion — soft, even, wraparound light; ideal for flat-lays and honest product color.
Add a direction and a quality — "soft window light from the left with a warm bounce fill" — and Hunyuan has a physically plausible target instead of a guess.
Why world knowledge helps (and how to use it)
Because Hunyuan Image 3.0 is a native multimodal model, it fills sparse prompts with accurate real-world detail. Name a real place, material, dish, or profession and it renders the plausible specifics — correct architecture, the right tools on a workbench, the right sheen on brushed steel — without you listing every element. That is a gift: it keeps prompts short and authentic.
The trick is knowing when to lean on it and when to override it. Lean on world knowledge for context and authenticity — the setting, the props, the cultural detail. Be explicit about the things you actually care about controlling: lighting, lens, composition, mood, and color. Short prompts resolve through world knowledge; long prompts (Hunyuan reads well over a thousand characters) let you pin down every specific when a shot has to land exactly.
10 photorealistic example prompts
Each of these follows the six-part formula. Paste one in, pick the ratio and 2K from the Settings line, then swap the bracketed or descriptive parts for your own subject.
1. Landscape — misty valley at dawn
A wide alpine valley at dawn, a river winding through pine forest below jagged peaks, thin mist pooling in the low ground and catching the first warm light spilling over the ridge, cool blue shadows against warm highlights, photorealistic landscape photography with sharp detail front to back and honest natural color, shot on a 24mm wide-angle lens at f8.Settings: Hunyuan Image 3.0 · 16:9 · 2K
Why: the 24mm at f8 keeps the whole scene crisp, and golden dawn light does the drama.
2. Environmental portrait — the blacksmith
A blacksmith in his fifties with a soot-streaked face and a heavy leather apron, glancing up from the anvil mid-strike, in a dim workshop with glowing forge coals behind him, warm low-key light with the forge as the key and deep shadows around, photorealistic documentary portrait with fine skin texture and honest color, shot on a 35mm lens at f2.Settings: Hunyuan Image 3.0 · 4:5 · 2K
Why: the forge as a motivated light source plus a 35mm reads as real reportage, not a studio setup.
3. Product macro — the wristwatch
A luxury automatic wristwatch with a deep-green sunburst dial and a brushed-steel bracelet, extreme macro three-quarter view showing the dial texture and beveled case, resting on a slab of dark slate, lit with soft grazing three-point light that catches the polished edges with one controlled reflection, premium photorealistic product photography with true metal tones, shot on a 100mm macro lens at f5.6.Settings: Hunyuan Image 3.0 · 3:2 · 2K
Why: the 100mm macro and grazing light sell the surface detail that makes premium products convincing.
4. Food — rustic breakfast
A rustic breakfast on a weathered wooden table, a plate of golden pancakes with dripping maple syrup, fresh berries, and a steaming cup of coffee, a linen napkin to one side, soft overcast window light from the left with gentle natural shadows, warm editorial food-photography style with rich texture and honest color, shallow focus, shot on a 50mm lens at f2.8.Settings: Hunyuan Image 3.0 · 4:5 · 2K
Why: soft window light plus a 50mm at f2.8 is the classic food-editorial recipe — appetizing without looking staged.
5. Interior — sunlit kitchen
A bright modern farmhouse kitchen with white shaker cabinets, a marble island, brass fixtures, and a bowl of lemons on the counter, warm morning sunlight streaming through a large window casting soft rectangles of light across the floor, airy natural palette, photorealistic interior photography with gentle depth of field and true color, shot on a 24mm lens at f5.6.Settings: Hunyuan Image 3.0 · 3:2 · 2K
Why: a 24mm opens up the room while the streaming morning light keeps it warm and believable.
6. Street — rainy crosswalk
A busy city crosswalk in the rain at dusk, pedestrians with umbrellas caught mid-stride, glowing shop signs and traffic lights reflecting off the wet asphalt, moody cinematic color grade with atmospheric haze and shallow focus on a woman in a red coat, photorealistic street reportage with subtle grain, shot on a 35mm lens at f1.8.Settings: Hunyuan Image 3.0 · 3:2 · 2K
Why: wet reflections and shallow focus at f1.8 give the candid, in-the-moment feel of real street photography.
7. Architecture — concert hall at blue hour
A sweeping modern concert hall with a curved glass-and-steel facade, low symmetrical wide composition, warm interior lights glowing through the glass against a deep-blue dusk sky, a few blurred figures at the entrance for scale, photorealistic architectural photography with crisp lines and true color, long-exposure feel, shot on a 24mm tilt-shift lens at f8.Settings: Hunyuan Image 3.0 · 16:9 · 2K
Why: a named tilt-shift lens keeps verticals straight, and blue hour balances warm interior glow against a rich sky.
8. Close-up texture — autumn leaf
An extreme close-up of a single frost-edged autumn maple leaf resting on dark wet bark, every vein and dew droplet sharp, rich red and amber tones against deep brown, soft diffused overcast light for even detail with no harsh glare, photorealistic macro nature photography with fine texture and natural color, shot on a 100mm macro lens at f4.Settings: Hunyuan Image 3.0 · 1:1 · 2K
Why: overcast diffusion plus a 100mm macro renders the veins and droplets without blown highlights.
9. Golden hour — fisherman on the dock
A weathered fisherman in a knit sweater mending nets on a wooden harbor dock at golden hour, moored boats and coiled ropes around him, warm low side light raking across his hands with long shadows and a soft rim on his hair, photorealistic documentary style with Kodak Portra tones and fine skin texture, shot on an 85mm lens at f2.Settings: Hunyuan Image 3.0 · 3:2 · 2K
Why: golden-hour side light, an 85mm, and a film-tone reference stack up to a warm, cinematic character portrait.
10. Blue hour — city skyline reflection
A city skyline at blue hour reflected in a calm river, glowing office windows and bridge lights mirrored on the water, a deep-blue gradient sky with the last warm glow on the horizon, still and atmospheric, photorealistic long-exposure cityscape with sharp detail and true color, deep depth of field, shot on a 35mm lens at f8.Settings: Hunyuan Image 3.0 · 16:9 · 2K
Why: blue hour plus a still-water reflection at f8 gives the balanced light and mirror symmetry that make cityscapes sing.
Mistakes to avoid
- Over-stuffing the prompt. Piling on twenty adjectives dilutes the ones that matter. Lead with subject, lighting, and lens; add detail only where the first result misses.
- Contradictory lighting. Asking for "golden hour" and "bright studio softbox" in the same prompt confuses the model. Pick one coherent setup and one direction.
- Saying "realistic" without specifics. The word alone does nothing. Realism comes from a named lens, aperture, light setup, honest color, and texture cues — not the adjective.
- Asking for 4K. Hunyuan Image 3.0 tops out at 2048x2048 (2K). Requesting 4K adds no detail and can muddy the description. Set the size to 2K in the tool.
- Ignoring lens and aperture. Without a focal length, you get an average, characterless look. Naming "50mm at f1.8" is the single fastest realism upgrade.
- Inventing flags. There is no --ar or --style parameter. Aspect ratio and resolution are interface or API settings; describe the ratio in plain words if you want, and set it in the tool.
- Fighting world knowledge. Do not spell out every generic detail Hunyuan already knows. Name the real thing, then spend your words on the choices you actually want to control.
Frequently Asked Questions
What is the formula for a photorealistic Hunyuan Image prompt?
Use six parts in order: subject and appearance, action or pose, environment or setting, lighting, camera and lens, then style and quality cues. For example: a weathered fisherman mending nets on a wooden dock at golden hour, warm low side light, shot on a 50mm lens at f2, photorealistic documentary style with fine skin texture. Naming a real lens, aperture, and named light setup is what pushes Hunyuan Image 3.0 from generic to convincingly photographic.
Why does naming a real lens and aperture improve realism?
A lens focal length and aperture encode a specific look the model has learned from real photographs. A 24mm gives sweeping wide-angle depth, an 85mm compresses and flatters faces, and a 100mm macro isolates fine texture. An aperture like f1.4 or f1.8 gives shallow depth of field with soft background blur, while f8 keeps a landscape sharp front to back. Naming these tells Hunyuan which photographic recipe to imitate instead of guessing.
What resolution can Hunyuan Image 3.0 output?
Hunyuan Image 3.0 generates up to 2048x2048, which is 2K. It is not a 4K model, so do not ask for 4K in your prompt — the request will not add detail and may confuse the description. Pick your aspect ratio (1:1, 4:3, 3:4, 3:2, 2:3, 16:9, or 9:16) and a size up to 2K in the interface or API. These are settings, not inline flags, so there is no --ar or resolution parameter to type in the prompt.
How does world knowledge help with photoreal prompts?
Because Hunyuan Image 3.0 is a native multimodal model, it fills sparse prompts with accurate real-world detail. Name a real place, material, dish, or profession and it renders the plausible specifics — the right architecture, the right tools, the right textures — without you spelling out every element. Lean on this for context and authenticity, then be explicit about the things you actually care about controlling, such as lighting, lens, composition, and mood.
What lighting terms make an image read as real?
Name a real setup rather than saying good lighting. Golden hour gives warm low side light and long shadows; soft window light is flattering and directional; three-point lighting is a clean studio standard; chiaroscuro and low-key give dramatic deep shadows; high-key gives bright, even, shadow-light; overcast gives soft, even diffusion. Pairing one named setup with a direction, such as soft window light from the left, gives Hunyuan a concrete, physically plausible target.
Should I write long or short prompts for photorealism?
Both work. Hunyuan Image 3.0 resolves short prompts through world knowledge, so a named subject and one light cue already produce a coherent photo. But it also understands very long prompts, well over a thousand characters, and honors the specifics. For controlled photoreal shots, write a full descriptive sentence covering all six formula parts; for quick exploration, start short and add detail only where the first result misses.
Why does asking for "realistic" not work on its own?
The word realistic is too vague to steer the model. Realism comes from concrete photographic choices: a named lens and aperture, a specific light setup and direction, honest color, and texture cues like fine skin detail or subtle film grain. Replace realistic with those specifics. Adding a film reference such as Kodak Portra tones or a shallow depth of field does far more for believability than the adjective alone.