Tools I use to get these results π
(Pro tip: If you want views like mine, the free tiers are just for testingβupgrade to the paid plans to get enough credits and maximum export quality)
First we have to create our assets.
To do that, we will work in GPT Image 2.0 or Seedream 5.0 in https://vyrl.pro/r/sickmanai and https://higgsfield.ai/?fpr=sickmanai

Input your selfie and paste this prompt:
SCENE CONTEXT
Character reference turnaround sheet of one young female knight in three full-body views (front, profile, back) plus one close frontal portrait, for identity and costume lock.
ACTIVE REFERENCES
<<<image_4>>> β identity reference (selfies, images 4β5): controls face, skin tone, eye color, facial proportions, front-view likeness. 100% matches reference.
<<<image_6>>> β identity reference (image 6): controls profile-only anatomy β nose bridge and tip shape, jaw and chin curve, lips seen in profile, ear shape, piercing placement. Used for Panel 2. Does not control hair length or hairstyle.
<<<image_1>>> (image 1) β primary reference for full costume: armor cut, engraving pattern, chainmail placement, leather belt rigging, cloak silhouette.
<<<image_3>>> (image 3) β reference for haircut shape only (bob length and cut line), not color.
CHARACTER ANCHORS
<<<image_4>>>: 20-year-old woman, 165cm tall, light slender build, no exaggerated bust or hip curvature.
Face shape: soft oval face, gently tapered toward a rounded chin, full cheeks with soft natural volume, no visible contouring or sharp angularity.
Skin: warm light-tan skin tone, smooth dewy complexion with natural visible skin texture, subtle natural sheen on the high points of the cheeks and nose bridge, no plastic or airbrushed finish.
Eyes: dark brown almond-shaped eyes, slightly hooded upper lid with a soft visible double-eyelid crease, long natural dark lashes, soft inner warmth in the iris, calm direct gaze.
Eyebrows: straight-to-softly-arched natural brows, medium thickness, dark brown, natural unshaped look, following the brow's natural growth line.
Nose: straight narrow bridge, soft rounded tip, small nostrils, straight profile line with no bump (per <<<image_6>>>).
Lips: full lips with a clearly defined cupid's bow, natural rose-pink tone with a soft glossy sheen, resting in a calm closed or very slightly parted expression.
Jaw and chin: softly rounded jawline curving into a gently pointed chin, no sharp angularity, smooth profile transition from ear to chin (per <<<image_6>>>).
Unique visible marks: small natural mole on the right cheek near the nose, small natural mole just below the left corner of the mouth β keep both moles visible and in the same position across all panels.
Ears: visible ear with one small cartilage stud piercing on the upper helix and two thin hoop earrings on the lower lobe, matching <<<image_6>>> exactly.
Face 100% matches the reference photos β front proportions from images 4β5, profile silhouette and ear detail from image 6.
HAIR SPECIFICATION
Blunt-cut jaw-length bob, cut roughly and unevenly at the ends, not salon-precise, no fringe, hair falls loose framing the face. Hair color stays her original dark brown-black from the selfie reference β do not use the blonde or auburn color from the armor reference images, and do not use the longer wavy length or center-parted fringe shown in image 6. Copy only the short bob length/shape from image 3, not its braid, wave, or color.
ARMOR AND COSTUME REFERENCE
Primary reference: image 1.
Materials: high-polish mirror-chrome steel plate armor, cool white-silver sheen (not warm gold, not matte pewter). Breastplate and pauldrons carry fine hand-engraved fleur-de-lis filigree along the raised edges. Riveted blackened steel chainmail fills the hip-skirt gap and underarm/inner-elbow joints. A brown leather sword-belt with aged buckles crosses the waist diagonally, holding a sheathed sword at the hip. Greaves, poleyns, and gauntlets are articulated silver plate with visible rivet lines at each joint, same mirror finish as the torso armor.
CLOAK
Full-length silk cloak in ivory-white, heavy satin sheen with visible woven texture, structured drape with real fabric weight, subtle self-tone brocade near the hem, clasped at the shoulders, moves as solid fabric mass, not thin or floating.
BODY AND PRESENTATION LOCK
No sexualized framing. No exaggerated bust, waist, or hip emphasis. Armor reads as functional plate protection with a straight, flat torso line. Slender 165cm frame, light build, modest and heroic presentation in every view.
SHEET LAYOUT
Single reference sheet, four panels on a neutral seamless light-grey studio background, identical flat lighting across all panels so armor material and color read true.
Panel 1 β full body, front view, neutral standing pose, arms relaxed at sides.
Panel 2 β full body, profile view (screen-left facing), nose/jaw/lips/ear silhouette locked to <<<image_6>>>, same pose and proportions as Panel 1.
Panel 3 β full body, back view, cloak visible in full length from behind.
Panel 4 β close portrait, direct frontal view, shoulders-up, face fully visible, calm neutral expression, eyes on camera, both moles and ear piercings visible.
LIGHTING
Soft even front-key studio lighting, low contrast, no strong directional shadow, no backlit silhouette. Priority is accurate material/color reproduction, consistent neutral daylight (~5600K) across all four panels.
POSITIVE CONSTRAINTS
Same face, same armor, same cloak, same hair length and color across all four panels.
Facial features (face shape, eyes, nose, lips, moles, ear piercings) identical and consistent in every panel.
Profile silhouette in Panel 2 matches <<<image_6>>> exactly.
Armor finish identical everywhere: cool white-silver mirror chrome, never gold or matte.
Hair color matches the original selfie (dark brown-black), never blonde/auburn, never long/wavy.
No added characters, no extra props, sword stays sheathed at hip.
No sexualized posing, no exaggerated body curves.
Face in Panel 4 fully visible, unobstructed by hair or shadow.
I used two different locations. One for the first scene and another for the second scene. I used the Higgsfield Soul Cinema image model because it gives a more cinematic and realistic look. https://higgsfield.ai/
.png)
LOCATION / ENVIRONMENT PROMPT β SCENE 1
A gigantic high-altitude medieval mountain cliff in a realistic fantasy world.
The location is a vast rocky mountain ridge overlooking an enormous medieval valley.
The cliff is covered with thick, naturally growing wild mountain grass, small patches of moss, scattered rocks and uneven weathered stone.
The terrain feels ancient, rugged and completely natural.
Beyond the cliff is a huge fog-filled valley.
Extremely dense cold mountain fog covers most of the valley, with only faint silhouettes of distant mountains occasionally visible through openings in the mist.
The fog is NOT static.
Large thick layers of mist continuously move rapidly through the valley and across the cliff.
Heavy mountain wind sweeps violently across the entire location.
The grass bends and ripples strongly in the wind.
Loose leaves, small pieces of vegetation and fine particles occasionally move through the air.
The wind pushes thick fog over the cliff edge and around the rocks.
The atmosphere feels extremely high-altitude, cold, damp and turbulent.
The sky is heavily overcast with soft diffused grey daylight.
No bright sunshine.
No dramatic sunset.
No clear blue sky.
Soft natural light diffuses through the thick fog.
The wet rocks have subtle realistic moisture and reflections.
The cliff should feel extremely high, with the valley below almost completely swallowed by fog.
The location must visually communicate:
AN ENORMOUS MOUNTAIN.
A MASSIVE CLIFF EDGE.
A VAST FOG-FILLED VALLEY.
EXTREME WIND.
DENSE MOVING FOG.
HIGH ALTITUDE.
AN OPEN SKY AHEAD.
The cliff edge should provide a clear, unobstructed open launch area.
There should be no buildings, castle, city or civilization visible in this scene.
Do NOT reveal the medieval city.
Do NOT reveal the villain.
Do NOT add characters.
Do NOT add animals.
Do NOT add creatures.
Do NOT add roads.
Do NOT add modern objects.
Do NOT add artificial structures.
The location should feel like a remote, untouched mountain overlooking the medieval world.
PHOTOREALISTIC LIVE-ACTION FANTASY MOVIE ENVIRONMENT.
Real rock formations.
Real grass.
Real atmospheric fog.
Real wind behavior.
Realistic atmospheric depth.
Natural weathering.
Natural terrain.
Natural lighting.
Extremely cinematic environmental scale.
The environment must remain consistent throughout the entire 10-second scene.
Strong wind should visibly disturb the environment.
The grass bends strongly.
Fog is pushed across and over the cliff.
Dense mist continuously swirls through the landscape.
The air feels turbulent.
The location should feel physically real, as if filmed on a gigantic mountain in a live-action fantasy movie.
No cartoon environment.
No video-game environment.
No artificial CGI look.
No fantasy illustration.
Full photorealism.