**Researched:** 2026-10-10. Sources marked `verified live` (read in the real browser) or `index` (search-index/fetch text). Confidence levels are noted per section.
---
**The original creator is the AI creator @sickmanai, whose character is called "Kitty"** — not @derek.mercer.2006 alone. The @derek.mercer.2006 connection is real but is a *collab/popularizer* role, per the timeline documented by makeviral.ai (which links each milestone to the original Instagram/X posts).
Timeline (all dates 2026):
| Date | Event |
|---|---|
| 19 Aug | @sickmanai posts part 1 of "Kitten vs Dragon" on Instagram: a fluffy kitten in silver armor and a pink bow sips an iced drink, then sets off to fight a dragon (478,000 likes by 7 Oct) |
| 21 Aug | Part 2: the kitten stands in a storm on a foggy field gripping a sword taller than itself — **this frame became the meme** (970,000 likes by 7 Oct) |
| 27 Sep | First big English post on X, "me trying to handle things maturely": 794,800 views, 23,000 likes |
| 28 Sep | **@sickmanai × @derek.mercer.2006 (AI gym-bro account) post "Remember your gym bro."** — a giant knight beside the tiny kitten; 3.1M Instagram likes and 3.4M TikTok plays by 7 Oct. This started the "tiny knight + big protector" format |
| 29 Sep – 5 Oct | Russian breakout: "котенок в доспехах" / "котенок с мечом"; Russian media explainers (T-J 2 Oct, then memepedia, vc.ru, DTF, Lifehacker) |
| 1 Oct | TikTok user casts their own cat, caption "remember your gym bro", 188,800+ plays |
| 3 Oct | @aimak.s posts "Little cat on the quest": white kitten knight vs black dragon, giant cat knight with crown steps in behind (462,000 likes by 9 Oct) |
| 4 Oct | @raizelfrmn posts "Bro trying to attack the sleeping dragon": kitten knight + tall knight in red cape charge a sleeping dragon (236,000 likes by 9 Oct); @sickmanai + @derek.mercer.2006 post "Me and Him" (250,000 likes); Google searches for "knight cat" hit their highest point so far; X runs a trending story on the "Kitten with Sword and Pink Bow" |
| 6 Oct | @khruromm posts "When she says, 'I am with you.'": kitten knight with pink bow beside a wounded knight in dark armor, then walking with him toward an army (118,000 likes by 9 Oct); creator Kevin Winnik posts a Higgsfield tutorial for the giant-and-tiny version, calling it the **"Muscular Character AI Video"** |
| 7 Oct | Parent-and-child versions: @gazizaaz's mom-and-toddler knight reel from Kazakhstan ("The daily fight for our children", sound: "Outro" by M83), then a dad-and-daughter clip on TikTok ("She's small. But she never stands alone.") |
**How @sickmanai made it (the only known breakdown of the original workflow):** Per the creator's own guide (cited by makeviral.ai; the creator sells the full long prompts through a guide and Patreon — original prompt text not publicly available):
· **Stills:** GPT Image 2.0 or Seedream 5.0
· **Video:** Seedance 2.5, animated in 10-second single shots
· Series format: many shots, full control; TikTok bio reads "Original creator of Kitty Knight"
No interview or behind-the-scenes video by @sickmanai was found in the searchable index; the guide/Patreon are paywalled.
---
**What it is:** a video-to-video flow (not a single model) that keeps motion, camera trajectory, VFX, masks/layer order, cut points, film grain and lens character from a reference clip while swapping the character/scene. Two modes: **Motion Transfer** and **Objects Swap**. Launched end of August 2026.
**Exact UI flow (official Higgsfield AI Influencer Studio blog, higgsfield.ai/blog/new-ai-influencer):**
1. Go to **higgsfield.ai/ai-influencer-studio** and select **Influencer builder**. The **Explore** tab shows finished characters and trending formats — selecting one applies its settings to the builder.
2. **Upload your photo** if you want one (optional — your own face or someone who gave permission, or pick one from your generations).
3. **Set the Character type:** Average, Bold, or Extreme (how far from ordinary the character reads).
4. **Work through the settings:** Identity, body, face, hair, style, accessories. **Randomize** fills everything at once; **Picked settings** shows the current configuration.
5. **Generate** — character generation costs 1.12 credits (~$0.06), shown on the button. The result lands in **History** as a portrait and a full-body shot.
6. **Review and reuse:** check the **Character Parameters** panel next to the result, rerun with **Recreate** if needed, and **save the character as an element** through the result menu.
7. **Move to video:** select **Turn to motion with Genjutsu**, pick **Motion transfer** or **Objects swap**, add a reference video, set the quality, and generate. The credit cost shows on the Generate button before you confirm.
**Genjutsu technical requirements/details (okihiro-school guide + detechtive.pl tutorial):**
· Reference video: 4–30 seconds, MP4 format.
· Use **@Image tags** to attach multiple reference images (character, product, background).
· Quality settings (720p/1080p); **Split View** to compare; you can split long videos to reduce credit usage.
· Tip from media.io's Genjutsu tutorial: use front-facing, full-body reference images with clear facial/costume detail and consistent, well-lit lighting when identity matters.
· Also inside Higgsfield: the **Motion Library** (gallery of ready-made motion templates) and **Recreate / Motion Control** let you start without writing a prompt from scratch.
**How it maps to knight cat specifically:** the @derek.mercer.2006 collab clip ("Remember your gym bro." — a gym-bro style video) serves as the reference video; a custom kitten-knight character replaces the person, keeping the original motion and camera move by move. (Mission premise: popularized by tutorial creator @stevenwommack — see caveat in section F.)
The classic pipeline used by the originator and copycats:
1. **Generate the knight still** in an image model: GPT Image 2.0 or Seedream 5.0 (creator's workflow); Nano Banana 2 (inside Higgsfield; also used by MakeViral's one-tap flow).
2. **Animate with image-to-video:** Seedance 2.5 (creator's choice, 10-second single shots); alternatives tested/mentioned: Seedance 2.0 Fast, Veo 3.1 Fast, Kling 3.0 (Standard option), Kling 2.6 (see B2b). Feed the still + a video prompt describing motion, camera, audio.
3. **Post:** add trending sound on-platform (see E), caption, hashtags (#knightcat, #catknight, #kittenwithsword, #kittenknight), keep the platform's AI label on.
**B2b. Kling 2.6 Motion Control (direct motion-transfer alternative, from TikTok creator @zhannashabal's tutorial caption):**
1. Upload a video with the dance (face emotions, gymnastics tricks, etc.).
2. Upload the character image (cat, yourself, friend, child, or any picture).
3. Launch video generation. Your character performs all actions from the original video.
· **MakeViral one-tap flow:** upload one full-body pet photo → Nano Banana 2 builds the knight still → Seedance 2.5 animates 10 s with sound → 9:16 MP4, no watermark. (`index`; makeviral.ai/templates/knight-cat-ai-video)
· **CapCut "Kitten With Sword" templates:** pick the template in the CapCut phone app, drop in a photo. Fastest route if you already edit in CapCut. (`index`, cited by makeviral.ai)
· **Reel-insertion (recast) flows:** H3 Max Recast or Genjutsu Motion put your pet into the *actual viral reels* shot by shot (12–15 s clips, keeping the reel's original song) — e.g. "I am with you" (@khruromm), "Dragon selfie" (@sickmanai), "Dragon quest" (@aimak.s), "Sleeping dragon" (@raizelfrmn). Prompts are short "replace character X with reference image" instructions (see prompt library).
---
> Note: the creator's (@sickmanai's) original prompts are paywalled (guide + Patreon) and were not found publicly. The long prompts below are MakeViral's own tested rewrites — the page explicitly states "we did not copy them and wrote and tested ours from scratch." Quoted verbatim as published.
> "Use the uploaded image as the base. Keep the animal's breed, face, eye color, fur color, fur length and markings exactly the same as in the photo. Turn the animal into a tiny, brave knight in a still from a real film. The animal stands upright on its hind legs and wears a full suit of fitted miniature medieval plate armor made for its small body: a polished, slightly scratched steel breastplate, rounded steel shoulder plates, jointed steel plates on the front legs, a short chainmail skirt, steel plates and small steel boots on the hind legs, leather straps with small brass buckles, and a short cream linen cape that whips sideways in the wind. No helmet: the head, face, ears and fur stay fully visible. With both front paws it grips the hilt of a real plain steel longsword at chest height and holds it straight up, point to the sky, just beside its face, so the blade never hides the face. The blade rises far above its ears, so the sword is about twice as tall as the animal. A strong wind blows from the left: the fur on its head and cheeks streams sideways. Its big eyes are open and look ahead, calm and brave. It is alone in the middle of a wide, misty moor of long, wet, green grass, with thick fog hiding the horizon. Overcast, soft, cool light, muted grey-green colors. Low camera at the animal's eye level, long lens, shallow depth of field, soft blurred grass in the foreground. Full body from ears to feet, with the whole sword in frame and open misty sky above. Photorealistic, natural film grain, not CGI, not a cartoon. Vertical 9:16. Exactly one animal, no people, no extra paws, no text or letters on the sword, no logos."
> "A slow, epic, cinematic scene on a misty moor, filmed like a real movie. Keep the animal's face, fur, markings and armor exactly as in the image. The little knight holds its ground in a strong, gusting wind: its fur streams sideways, the short cream cape snaps and flutters, the long grass ripples in waves and mist drifts past. For the first seconds it stands perfectly still and calm, gripping the tall sword. Then a huge gust hits: it narrows its eyes, grips the sword tighter with both front paws, plants its feet and leans into the wind without taking a step back. As the gust passes, it slowly lifts its chin and looks bravely ahead into the fog. The sword stays straight up the whole time. The camera stays low and close at the animal's eye level with a gentle handheld drift and a very slow push in, so the little knight stays big in the frame the whole time. Natural, real animal motion. Audio: loud howling wind and the cape flapping from the first second, soft clinks of armor, then an original slow, epic orchestral score with deep drums and strings that swells when the big gust hits. No people, no extra paws, no morphing, no text, no logos."
> "Use the two uploaded images as the base. The person in the first image is the big knight. The animal in the second image is the little knight. Keep the person's face, facial features, skin tone, hair and body shape exactly the same as in their photo. Keep the animal's breed, face, eye color, fur color and markings exactly the same as in its photo. Do not blend the person and the animal. A still from a real film on a wide, misty moor of long, wet, green grass, with thick fog hiding the horizon, overcast soft cool light and a strong wind from the left. In the front, the animal stands upright on its hind legs in a full suit of fitted miniature medieval plate armor made for its small body: a polished steel breastplate, rounded steel shoulder plates, jointed plates on the front legs, a short chainmail skirt, steel plates and small steel boots on the hind legs, leather straps, small brass buckles and a short cream cape that whips in the wind. No helmet on the animal. With both front paws it grips a real plain steel longsword at chest height and holds it straight up, point to the sky, just beside its face; the blade rises far above its ears. A few steps behind the animal and to the right, the person stands tall in full polished steel plate armor with no helmet and a long dark red cape with gold trim that streams in the wind. The person holds a longsword point-down at their side and looks down at the little knight with a calm, proud face and a faint, gentle smile. Low camera at the animal's eye level. The animal is full body from ears to feet, the person is full body from head to feet, and both faces are clear and sharp. Photorealistic, natural film grain, muted grey-green colors, not CGI. Vertical 9:16. Exactly one person and one animal, no extra paws, no extra limbs, no text or letters on the swords, no logos."
> "A slow, epic, cinematic scene on a misty moor in a strong wind, filmed like a real movie. Keep the person's face, hair and armor and the animal's face, fur and markings exactly as in the image. The little animal knight stands in front and holds its ground, gripping the tall sword while its fur and short cape stream in the wind. Behind it, the person in plate armor and a long dark red cape walks slowly forward through the wind, kneels on one knee right beside the little knight, plants their sword point-down in the grass, and rests one armored hand gently on the animal's back to protect it. The animal looks up at the person, and the person looks back with a soft, proud smile. Then both turn their heads and look out to the left into the wind with calm, brave faces, side by side, standing guard, while the long red cape and the fur whip in the wind and mist drifts past. Their faces stay visible in three-quarter view, never turned away from the camera. One continuous shot with no cuts. The camera stays low and almost still, with a slight handheld sway. Natural motion. Audio: howling wind from the first second, the cape flapping, armor clinking with each step, and an original slow, emotional orchestral score with sweeping strings and deep drums that swells when the person kneels. No dialogue, no lyrics. Exactly one person and one animal. No smoke, no fire, no extra limbs or paws, no merged bodies, no morphing, no text, no logos."
**"I am with you" (@khruromm reel; H3 Max Recast, 12 s, 720p):**
> "Image 1 replaces the little kitten knight with the pink bow and the cream cape, who stands beside the big knight and then walks with him toward the army, in every shot, with no bow. Keep the big knight in dark armor with the closed helmet, the arrow in his shoulder, the bloody sword, the army, the battlefield, the motion and the camera exactly as in the video. Keep the white caption text exactly as in the video."
**"Sleeping dragon" (@raizelfrmn reel; Genjutsu Motion, 14 s, 480p):**
> "Replace the tall knight in dark armor and the long red cape with the person from image 1: keep that face, hair and skin tone exactly, with no helmet. Replace the tiny kitten knight that holds the sword with the pet from image 2: keep its face, fur colors and markings exactly, with no bow. Both swords are plain steel, not glowing. Keep the armor, the capes, the moor, the sleeping dragon, the motion and the camera of the video."
**"Dragon quest" (@aimak.s reel; H3 Max Recast, 15 s, 720p, 16:9):**
> "Image 1 replaces the little white kitten knight with the sword and the short cape, in every shot, with no bow and no ribbon. Keep the giant black cat knight with the crown, the black dragon, the lightning, the background, the motion and the camera exactly as in the video."
**"Dragon selfie" (@sickmanai reel; H3 Max Recast, 15 s, 720p):**
> "Image 1 replaces the kitten knight with the pink bow who sips the iced drink and then stands up to the dragon, in every shot, with no bow. Keep the red dragon, the field, the fog, the motion and the camera exactly as in the video."
· "person frozen in time, only camera perspective changes, whipping around the persons back in a circular motion 180°." (Kling-powered Strava edit tutorial; 135.8K likes)
· "transform background. Main character does not move or change position. Emotionless. Only background elements transform." (Kling 2.5 Turbo inside Higgsfield AI)
· Style: very short, declarative, subject-verb motion directives; works inside Higgsfield which hosts Kling models.
· Openjourney 4 (PromptHero): "An English short cat wearing a medieval knightly dress,dark background, gothic, insanely detailed, 8K picture quality, movie lighting, powerful picture sense"
· SDXL 1.0 (NightCafe, 2024): "Chibi knight cat character in shining armor, tiny sword, large helmet with feather, valiant pose, medieval castle background, 3D model style, modeling clay, polymer clay, large dot eyes, simple cute features, heraldic banner, inspired by Japanese kawaii design, photorealistic, high resolution"
· Tripo H3.1 (3D): "cat knight figurine with purple sword, armored cat in plate armor, brown fur, pink bow accessory, standing on a round base"
· yesand.ai (ex-Inspova) hosts "Knight Cat Challenge" remix entries ("Knight Cat Fought a Dragon" — source: Sickman; "Knight Kitten stands in the wind") but prompt text is JS-rendered and not readable in text fetch.
---
1. **Higgsfield saved character elements:** build the character once in AI Influencer Studio, check the Character Parameters panel, save it as an **element** via the result menu, then reuse it in Genjutsu Motion Transfer / Objects Swap — the motion changes but the character identity persists. (higgsfield.ai official blog)
2. **Reference-image recast prompts:** attach the pet's photo with **@Image tags** and write prompts that lock identity: "Keep the animal's breed, face, eye color, fur color, fur length and markings exactly the same as in the photo." (okihiro-school; MakeViral)
3. **@stevenwommack-style minimal motion directives:** describe only the motion, never redescribe the character — "Main character does not move or change position… only background elements transform."
4. **Nano Banana (Gemini image) native consistency:** upload a clear, front-facing reference; refer to it as "this character"; state what must stay identical; generate a series one image at a time **in the same chat** so context persists; re-upload the original reference at the first sign of drift; Nano Banana Pro supports up to 14 reference images (6 high-fidelity) and multi-image fusion (character + outfit + background). (flick.art; chatsmith.io; help.apiyi.com)
5. **Identity-lock instruction line:** append to every frame prompt: "Keep the character's facial features exactly the same as the reference image. Maintain identical attire and hairstyle throughout." (help.apiyi.com)
6. **Fixed seed:** lock the random seed across generations (e.g. `extra_body={"seed": 42}` in the Gemini image API) to hold style/appearance stable. (help.apiyi.com; OpenArt shows seed per creation)
7. **Character sheet method:** generate front, 3/4-side, and back views on a clean background first; chain images (A→B→C) to build character data before producing final shots. (lilys.ai; help.apiyi.com)
8. **Short reference clips:** if identities drift in motion transfer, shorten the reference clip and simplify the cast design; refine one variable at a time. (media.io)
---
| Method | Model / route | Cost cited |
|---|---|---|
| Higgsfield character build | AI Influencer Studio | 1.12 credits (~$0.06) |
| Higgsfield Genjutsu generation | 720p / higher quality | 65 credits at 720p, up to 90 higher (detechtive.pl, 2026); cost shown on Generate button |
| Higgsfield plans | — | Starter $15/mo, Plus $49/mo ($39/yr), Ultra $129/mo ($99/yr); 20 credits per $1 (official blog, Oct 2026). Promo seen: 70% off plan + 7 days unlimited generations on select models |
| MakeViral knight-cat one-tap | Nano Banana 2 still + Seedance 2.5 (10 s, 720p, 9:16) | 128 credits/video; smallest one-time pack covering it: $14.99 / 150 credits, no subscription |
| By hand (MakeViral AI Video) | Seedance 2.5 (10 s 720p + 1 photo) | 128 credits |
| By hand | Seedance 2.0 Fast (10 s 720p + 1 photo) | 68 credits |
| By hand | Veo 3.1 Fast (8 s 720p + 1 photo) | 14 credits |
| By hand | Kling 3.0 (Standard) | cheaper option on baby-knight page; Best = Seedance 2.5 |
| Reel recast | H3 Max Recast (12–15 s, 720p) | 144–180 credits |
| Reel recast | Genjutsu Motion (14 s, 480p) | 179 credits |
| Creator's original toolchain | GPT Image 2.0 or Seedream 5.0 → Seedance 2.5 | not publicly priced in sources found |
**Sounds people use (for posting):** "worry (ultra slowed)" by LONOWN and riserayss (most reused named TikTok sound in the trend); "Outro" by M83 (viral mom/baby version); "Army Dreamers" by Kate Bush, slowed. Proven captions: "me trying to handle things maturely", "Me and Him", "remember your gym bro".
---
| Tutorial | Key steps |
|---|---|
| **Kevin Winnik — Higgsfield tutorial, "Muscular Character AI Video" (6 Oct 2026, Instagram)** | Taught the giant-and-tiny knight-cat version in Higgsfield; the tutorial that spread the owner-and-pet format. (Full step text not retrievable via index; cited by makeviral.ai) |
| **detechtive.pl — "I put my face into a movie scene (Genjutsu AI)" (YouTube; sponsored)** https://www.youtube.com/watch?v=zqKrTPLVZM0 | Reference photos + prompt → Motion Transfer / Object Swap; Gallery of ready-mades, Recreate, Motion Control; demo: Gemini image → animate in Seedance 2.5 → replace in Genjutsu; cost 65 credits @720p / up to 90 higher; free queue mode |
| **"How to Use Higgsfield Genjutsu | Keep the Motion, Replace the Person and Background" (YouTube)** https://www.youtube.com/watch?v=jJls9eFdWnw | @Image tags for multiple references; video requirements 4–30 s MP4; quality settings; Split View; splitting videos to save credits; full guide at okihiro-school.com |
| **"Higgsfield Genjutsu: One Video, Every Version You Need" (YouTube)** https://www.youtube.com/watch?v=0moHSN0-L5E | Walkthrough of higgsfield.ai workflow live: Motion transfer, Objects swap, Motion Library, uploads + quality settings, credits, real generation wait; pros/cons |
| **@zhannashabal — Kling 2.6 Motion Control tutorial (TikTok)** https://www.tiktok.com/@zhannashabal/video/7594818674557488395 | 1) upload dance/reference video 2) upload character image 3) generate — character performs the original video's actions |
| **@stevenwommack — Higgsfield/Kling prompt tutorials (TikTok, 2.6M followers)** https://www.tiktok.com/@stevenwommack | 135.8K-like Strava edit tutorial with verbatim prompt; Kling 2.5 Turbo-in-Higgsfield background-transform tutorial; Nano Banana 2-in-Higgsfield image tutorials |
| **Higgsfield official — "Introducing the New AI Influencer" (blog)** https://higgsfield.ai/blog/new-ai-influencer | 7-step character builder → Genjutsu motion flow (quoted in B1) |
| **media.io — "Higgsfield AI Genjutsu Tutorial"** https://www.media.io/video-effects/higgsfield-ai-genjutsu-tutorial.html | Upload video + references → describe only the change → generate preview → check identity/contact points/occlusion → refine one variable at a time |
---
1. **@sickmanai's actual original prompts** — paywalled behind his guide + Patreon; no interview or BTS found in the index. Only his toolchain is known (GPT Image 2.0 or Seedream 5.0 → Seedance 2.5, 10 s single shots).
2. **@stevenwommack's specific knight-cat Genjutsu tutorial** — his account is verified live (2.6M followers, bio "Mobile Content Tips and Tricks", major Higgsfield tutorial creator), and his verbatim prompt style is documented from other tutorials, but the index contained no retrievable knight-cat-specific tutorial of his; TikTok live search rendered no results (login/bot wall). The claim that he popularized the knight-cat Genjutsu method rests on the mission brief, not an independently found source.
3. **Kevin Winnik's 6 Oct tutorial steps** — only second-hand description found (makeviral.ai); the Instagram post itself wasn't retrievable via text fetch.
4. **TikTok caption-level prompts** — TikTok pages are largely JS/login-gated; prompts in captions/comments couldn't be extracted.
5. The mission's premise that @derek.mercer.2006 is *the* originator is partially corrected by sources: he co-posted the breakout collab ("Remember your gym bro.", 28 Sep) with @sickmanai; the original "Kitty" character and first two reels are @sickmanai's alone (19/21 Aug).
No page instructions were encountered that attempted to redirect the task; TikTok's login wall was recorded and disclosed rather than bypassed.
· https://www.makeviral.ai/templates/knight-cat-ai-video — trend timeline, original workflow, all C1–C5 prompts, costs, sounds, troubleshooting (`index`, read 2026-10-10)
· https://www.makeviral.ai/templates/baby-knight-ai-video — origin timeline recap, Kling 3.0 vs Seedance 2.5 cost note, parent/child variant (`index`, read 2026-10-10)
· https://higgsfield.ai/blog/new-ai-influencer — official AI Influencer/Genjutsu UI steps + pricing (`index`, read 2026-10-10)
· https://www.youtube.com/watch?v=zqKrTPLVZM0 — detechtive.pl Genjutsu tutorial, credit costs (`index`, read 2026-10-10)
· https://www.youtube.com/watch?v=jJls9eFdWnw — Genjutsu how-to, @Image tags, 4–30 s MP4, Split View (`index`, read 2026-10-10)
· https://www.youtube.com/watch?v=0moHSN0-L5E — live higgsfield.ai Genjutsu walkthrough (`index`, read 2026-10-10)
· https://www.media.io/video-effects/higgsfield-ai-genjutsu-tutorial.html — Genjutsu reference/prompt discipline (`index`, read 2026-10-10)
· https://kannadabindu.in/how-to-create-trending-higgsfield-ai-motion-transfer-videos-using-prompts-in-2026/ — motion-transfer steps (`index`, read 2026-10-10)
· https://www.tiktok.com/@stevenwommack — profile `verified live` 2026-10-10 (2.6M followers); tutorial captions/prompts via search snippets (`index`)
· https://www.tiktok.com/@zhannashabal/video/7594818674557488395 — Kling 2.6 Motion Control steps (`index`)
· https://yesand.ai/prompt/knight-cat-fought-a-dragon-77276c5b and https://yesand.ai/prompt/knight-kitten-stands-in-the-wind-ac387180 — Knight Cat Challenge remix community (prompt text JS-gated) (`index`)
· https://help.apiyi.com/en/nano-banana-pro-ai-video-storyboard-character-consistency-guide-en.html — consistency techniques, seed locking (`index`)
· https://flick.art/blog/img2img-consistent-character/nano-banana — 5-step consistency flow (`index`)
· https://chatsmith.io/blogs/ai-guide/how-to-use-gemini-nano-banana-00033 — reference-based consistency (`index`)
· https://lilys.ai/en/notes/nano-banana-20251026/fix-nano-banana-biggest-problem — chained character-data method (`index`)
· https://www.eweek.com/news/google-nano-banana-2-cheat-sheet/ — consistency prompt starter (`index`)
· https://prompthero.com/prompt/16c3a87fc2a, https://creator.nightcafe.studio/creation/7nNW7aiUsRLkUPwIbhrQ, https://studio.tripo3d.ai/3d-model/cat-knight-figurine-with-purple-sword-armored-cat-in-plate-armor-bro-e856c941-299c-4fc8-8a89-28397e7943a7 — prompt-sharing examples (`index`)
· https://www.tiktok.com/search?q=stevenwommack%20knight%20cat%20genjutsu — live check: no results rendered (bot/login wall) (`verified live` 2026-10-10)