Issue #18 | Platform: GPT‑Images 2 + Nano Banana 2 | July 14, 2026
"A single word that encodes identity or story is worth more than fifty adjectives of style. Models follow structure, not vibes."
[02] THE FIRST 80 WORDS
Every high‑demand AI image prompt that fails now does so for the same reason. The user describes aesthetics, not identity. They ask for “cinematic cyberpunk portrait” or “viral AI art” and skip who the subject is, what story the image should tell, and which feed it has to survive in.ltx+2
The market has shifted. People are paying for prompts that encode persona, platform, and outcome. Identity‑driven portraits, nostalgic edits and short vertical video hooks that feel feed‑native are what GPT‑Images 2, Nano Banana and their peers are actually being used for in 2026.cliprise+3
This issue gives you a single cross‑platform structure for identity and story, then shows you exactly how it translates into both images and clips.
[03] PLATFORM SPOTLIGHT. Nano Banana 2
GPT‑Images 2 and Nano Banana 2 | 2026 behavior
Across GPT‑Images 2 and Nano Banana 2, the strongest demand is not for abstract art styles. It is for prompts that produce identity‑anchored portraits, branded content and scroll‑stopping assets that fit Instagram, TikTok and Shorts without extra rework.cliprise+3
On the image side, users lean on GPT‑Images 2 and Nano Banana 2 for personalized photo editing and creator branding: scrapbook overlays, childhood‑self reworks, caricature‑style portraits, “ideal partner” couple visuals and editorial magazine‑style shots that feel like a specific person’s feed rather than a generic moodboard.piclumen+3
On the video side, the same audience turns to models like Sora, Kling and Veo for highly cinematic, 4K, vertical clips with clear motion cues. The winning prompts mix identity with action: “young tech founder walking through neon city at night” or “travel creator filming market scenes in Marrakech” rather than just “cyberpunk city in motion”.cliprise+3
The surprise for most users is that these models respond best when identity, platform and outcome are spelled out in short, structured segments, not buried in adjective lists. That is where you get production‑ready layouts, legible text, consistent characters and clips that feel designed for a specific social slot.
INCOME SIGNAL:
Identity‑driven prompt systems are already being sold as products: prompt libraries and packs for “portrait storytelling”, “creator branding”, “product hero shots” and “faceless channel hooks” across GPT, Gemini, Midjourney, Sora and Kling. If you can turn a single identity‑plus‑platform master prompt into usable image and video outputs for creators, there is a direct pathway to subscription libraries, freelance packages and training offers.
[04] THE TECHNIQUE: NAME THE IDENTITY FRAME, NOT JUST THE STYLE DIAGNOSIS
DIAGNOSIS
The failure pattern is now predictable. A user writes: “cyberpunk portrait, Instagram style, cinematic lighting, trending, high quality.” The model produces an image. It looks impressive. It also looks like every other AI portrait on the explore page because the prompt activated style clusters, not a specific identity or narrative.
They did not make a technical mistake. They described aesthetics. The model needs an identity frame.
MECHANISM
Modern image and video models learn both style and subject clusters. “Neon noir cityscape” is a style bucket. “First‑time founder at his desk at midnight, tired but determined, lit by laptop glow” is an identity and story bucket, which pulls different training examples and composition decisions. AI image generation trend 2026
When you name only style, the model blends across anonymous subjects and produces a plausible average. When you name identity, role, emotional state and platform slot, it picks a narrower cluster and renders decisions that show up as specific clothing, posture, eye contact and framing. Read more here…
The result of naming identity is differentiation. The result of describing style alone is algorithmically familiar output that competes poorly in feeds.
[05] THE STRUCTURE - WHERE THIS TECHNIQUE LIVES
This technique lives in the Subject and Context part of your prompt structure, but it changes what happens in Platform and Technical Parameters too.
Subject and Identity
Name the role, the emotional state and the long‑term story.
“Corporate female executive updating her LinkedIn profile after a promotion.”
“Emerging creator redesigning her Instagram grid to feel more ‘main character’ and less generic.”
Identity forces choices about pose, wardrobe and expression before you ever mention style.
Platform Context
Attach a target slot: “LinkedIn profile photo”, “TikTok 9:16 short”, “YouTube thumbnail”, “Instagram carousel panel 1”. Models like GPT‑Images 2, Nano Banana and video engines respond better when they know whether the output is a single portrait, a scroll‑stopping cover, or a moving hook.
Technical Parameters
On images, keep the lighting and focal length logic from your previous issue. On video, extend it into camera motion and duration.
For portraits: Rembrandt lighting, 85mm, shallow depth of field, legible text region if needed.
For clips: 4K, 9:16, three‑second lead‑in, one clear camera move (dolly, orbit, follow) and one satisfying motion (hair, fabric, product, water).
BEFORE / AFTER
BEFORE:
Cinematic AI portrait of a young creator. Neon lighting. Trending. High quality.
AFTER:
Portrait of a young female creator in her mid‑20s, primary platform TikTok, known for cozy night‑routine content. In‑bedroom scene at 11:30pm, fairy lights and laptop glow. Identity frame: “building a career from her bedroom, exhausted but hopeful”. Rembrandt lighting from laptop screen and bedside lamp, 85mm portrait lens feel, shallow depth of field. Composition framed for a vertical TikTok profile photo crop, with space for overlay text above the head. Skin and hair rendered authentically with visible texture, no beauty smoothing, slight under‑eye shadows. Color palette: warm amber and muted teal for feed cohesion.
WHAT TO WATCH FOR
Identity line clarity.
On identity‑first outputs, the persona line should be visible in the image or clip. If you prompt “emerging TikTok creator rebuilding her grid” and the result looks like a generic fashion model or abstract art, the identity frame is too vague. Replace broad labels (“young creator”) with concrete roles, contexts and emotional states (“night‑routine creator filming in her bedroom after a long shift”).Platform framing.
In a correctly framed prompt, composition and aspect ratio match the platform slot you named. A “LinkedIn profile photo” should read as a clean, centered portrait with minimal clutter. A “TikTok 9:16 hook” should read as a vertical clip with a strong focal subject and motion in the first seconds. If your “for TikTok” outputs crop like 16:9 B‑roll or your “for LinkedIn” portraits look like cinematic film stills, the platform line is not doing any work. Move it to the top of the prompt and make it explicit (“TikTok 9:16 short”, “LinkedIn profile headshot”).Story consistency across runs.
When the identity frame is doing its job, multiple generations feel like variations of the same person’s world, not entirely different characters. Wardrobe, environment class, color palette and emotional tone should stay recognizably within the same narrative, even as you change lighting or camera moves. If every rerun looks like a new persona, your identity block is too short or too abstract. Expand it into one clear sentence of long‑term story (“building a career from her bedroom”, “relaunching her brand after a setback”) and keep that line unchanged across tests.
INCOME APPLICATION
Identity‑driven portraits and short vertical hooks are already a direct, repeatable service category. Creators and professionals pay for assets that look like them and fit their primary platform, not just “good AI art”.
A single identity‑first prompt structure, run across a few archetypes, produces a reference portfolio. That portfolio is your proof of service.
Three archetypes. Corporate authority (LinkedIn executive), creative director (Instagram grid and portfolio), technical professional or founder (LinkedIn plus launch page). Each gets a portrait and a matching short video hook built from the same identity frame.
Twelve‑asset package. For each archetype, deliver one primary portrait, one vertical hook clip, and two derivative assets (thumbnail crops or alternate platform versions). That gives you a twelve‑asset selection package that demonstrates how identity‑first prompting travels across image and video.
Second income pathway. Still portraits and brand‑neutral identity frames can be repurposed into stock and template products, as long as you respect disclosure and release requirements on platforms like Adobe Stock and use generic, non‑real identities for public libraries.
This is the same logic as your lighting issue. There, one named setup across multiple subjects created a portfolio for headshot work. Here, one identity‑first structure across multiple platforms creates a portfolio for persona‑based AI branding services that can evolve into prompt packs, libraries and training offers.
[06] PROMPT VAULT ENTRY
PROMPT VAULT ENTRY
VAULT ID: PTP‑ID‑IMGVID‑016.1.0
PLATFORMS: GPT‑Images 2 + Nano Banana 2 + any video model with 4K 9:16 output
TECHNIQUE: Identity‑First Feed Prompt Structure
QUALITY SCORE: 23/25
TAGS: identity, portrait, vertical video, creator branding, commercial, GPT‑Images 2, Nano Banana 2
INCOME TIER: COMMERCIAL
IMAGE MASTER PROMPT (GPT‑Images 2 / Nano Banana 2):
Professional social portrait of [SUBJECT_IDENTITY], captured for [PLATFORM_SLOT].
Subject: [ROLE], [AGE_RANGE], [EMOTION_STATE], wardrobe [WARDROBE_DESCRIPTION] matching [COLOR_PALETTE] brand colors.
Identity frame: [LONG_TERM_STORY] expressed through pose, expression and environment.
Lighting: [NAMED_SETUP] lighting, [LIGHTING_DETAIL] descriptor, [FILL_RATIO] ratio.
Lens and framing: [FOCAL_LENGTH] portrait lens equivalent, shallow depth of field, composition centered for [ASPECT_RATIO] crop with clear space for overlay text in [TEXT_REGION].
Background: [ENVIRONMENT_SCENE] matching subject’s real content niche, blurred but recognizable.
Skin and texture: authentic rendering with visible pores and hair texture, no aggressive beauty filter, slight asymmetry preserved.
Resolution: 4K, tuned for social feed export.
VIDEO MASTER PROMPT (Sora / Kling / Veo):
4‑second vertical short, 9:16, featuring [SUBJECT_IDENTITY] in [ENVIRONMENT_SCENE] acting within [LONG_TERM_STORY].
Camera: [CAMERA_MOVE] motion, starting on a tight shot then [SECONDARY_MOVE] to reveal context.
Motion cue: one satisfying movement tied to identity, such as [MOTION_ELEMENT] that repeats across this creator’s content.
Lighting and color: match the still portrait look, [NAMED_SETUP] directional lighting, [COLOR_PALETTE] grade for brand cohesion.
Audio: ambient sound and subtle underscore matching [EMOTION_STATE], no voiceover.
Export: 4K, 24fps, optimized for TikTok / Shorts hook.
VARIABLES:
[SUBJECT_IDENTITY]: “corporate female executive updating LinkedIn after promotion”, “male tech founder launching a new product”, “creative director refreshing her portfolio grid”.
[PLATFORM_SLOT]: LinkedIn profile photo, TikTok hook frame, YouTube thumbnail, Instagram carousel panel.
[NAMED_SETUP]: Rembrandt, Butterfly, Loop, Split, depending on desired authority or approachability.
[FOCAL_LENGTH]: 85mm for authority, 50mm for environmental storytelling, 135mm for glamorized editorial.
USAGE NOTES:
Treat this as the base prompt for identity‑driven portraits and matching video hooks. Run three subject identities before client delivery to confirm that the persona reads clearly in both still and motion formats. Store the best‑performing combinations as named templates in your private prompt library instead of keeping them as one‑off prompts.
INCOME PATHWAY NOTE:
This dual image‑video prompt structure produces assets suitable for creator brand kits, LinkedIn and Instagram profile upgrades, visual sales pages, and faceless channel intros.
Combined with current Adobe Stock and platform‑safe policies, the stills can be repurposed into stock submissions with appropriate disclosure and releases, while the clips become paid deliverables for social campaigns.SCORING BREAKDOWN:
Clarity 5/5, identity and platform slots are explicitly named.
Output Control 5/5, constrains to specific formats and feed roles.
Reasoning – 5/5, directs the model’s interpretation from identity down to lighting and motion.
Hallucination Defense 4/5, real‑world identity details still need verification and client approval. Add “do not invent biographical details” when necessary.
Token Efficiency 4/5, some environmental descriptors can be compressed when you lock into a niche.
TOTAL: 23/25, above the 18/25 vault threshold.
[07] PLATFORM COMPARISON SNAPSHOT
Platform-honest bottom line: Identity‑first prompting is portable across GPT‑Images 2, Nano Banana and video engines, but the details are not. Image prompts are more model‑specific than chat prompts, so your best strategy is to build a small private library per platform rather than chasing 10,000 generic prompts that never mention who the subject actually is.
[08] INCOME SPOTLIGHT CALLOUT
THE INCOME SIGNAL:
The ability to name a persona and platform slot in a prompt is the difference between generating a pretty picture and generating a billable brand asset. “Rembrandt” can turn a generic headshot into a professional portrait. “First‑time founder updating LinkedIn after promotion” can turn that portrait into a narrative asset someone keeps for three years instead of replacing in six months.
THE MINIMUM VIABLE WORKFLOW
Take the master identity‑first prompt from Vault ID PTP‑ID‑IMGVID‑016.1.0.
Run three subject identities: corporate female executive, male tech founder, creative director, each with a clear platform slot.
Generate a still portrait set plus one short vertical video hook per identity using the same persona frame.
Deliver a twelve‑asset package: three portraits, three clips, six derivative thumbnails or crops for different platforms.
THE PLATFORM MATCH
GPT‑Images 2 and Nano Banana 2 for fast, identity‑anchored portraits, thumbnails and grid assets. Sora, Kling or Veo for matching short clips that inherit the same persona and color system. Prompt libraries like YouMind or Sozee for scaling this workflow into reusable templates you can sell or teach. gemini
THE STARTING POINT
Your first product does not have to be a massive library. A focused “Identity Starter Pack” for GPT‑Images 2 and Nano Banana, with three personas, three platform slots and three matching video hooks, is enough to prove value. Pair it with a short Loom or Gemini‑generated tutorial and you have a viable digital product for creators who are already searching for structured, reusable prompts instead of one‑off magic phrases. Check the 2026 best ai prompt library.
[09] YOUR ASSIGNMENT BEFORE THE NEXT ISSUE
THE TASK
Pick one identity frame from this issue. For example, “corporate female executive updating her LinkedIn profile after a promotion” or “emerging TikTok creator rebuilding her grid to feel more main character”. Use the master vault prompt as your base.
Generate three portraits and three short clips of the same subject. For the portraits, change only the named lighting setup and platform slot. For the clips, change only the environment and camera move. Keep identity, emotion and color palette identical.
THE SUCCESS CONDITION
The portraits show a visible, measurable difference in authority and approachability between setups. The clips show clear changes in motion and emotional tone while still feeling like the same person’s brand. If all six outputs feel interchangeable, the identity frame was too generic or the platform slot too vague.
THE FAILURE SIGNAL
All portraits look like stock AI headshots. All clips look like generic B‑roll. That is the sign that your prompt is still describing style, not identity. Move the persona line to the very top of the prompt, explicitly name the platform slot, and reduce the adjective stack until the model has to pay attention to who this is and what story they are in.
THE INCOME TEST
If one portrait and one clip look like assets you would pay for as a creator or marketer, you have a proof‑of‑service pair. Screenshot both. Put them in a folder called “Identity Kit Portfolio”. They are the first slides in your pitch deck for identity‑driven AI visual services.