
Image Director Kit
Format-first prompts, scored renders and on-format series, so AI images look finished.
GetKits 제작
$39의 가치가 있는 이유
대신 맡아 주는 일
About 30 minutes of your week: the Bot turns Monday's briefs into prompts at 9:00 AM and scores Friday's renders at 2:00 PM, the Skill revises in your chat, and your dot keeps each series on format.
첫날부터 바로 쓰는 것
- The Skill, the Bot and the Dot, sharing 24 formats and one rubric
- A playbook: setup order and how a render moves from brief to keeper
- About 30 minutes to install, plus about 1 hour for a series sheet
- 4 metrics with targets and a week-by-week plan for the first 30 days
직접 만들지 않아도 되는 것
Getting three AI tools to share one format library, one 30-point rubric and one naming scheme, so a render the Bot scores is revised by the Skill without translation. The playbook is the part no single piece includes.
이 Kit 3개를 따로 사면 $57입니다. 묶음은 $39이라 $18 절약됩니다.
세 가지가 함께 돌아가는 모습
요청하면 Skill이 일을 처리하고, Bot이 Grok에서 루틴을 돌리고, Dot이 ChatGPT에서 계속 지켜봐요. 아래 각 단계는 이 Kit에 들어 있는 실제 전후 비교예요.
사용 전
My Midjourney prompt: "silver-haired cyberpunk girl, neon city, 8K, ultra detailed, cinematic lighting, masterpiece". Every render looks like every other AI image. What am I missing?
사용 후
Score: 9/30 Verdict: Test renderMissing: the object. She floats in neon haze with no edge, no job and no metadata.Tells: void background, ID-photo framing, quality incantations, rim light on everything.1. Open with the format: "The front cover of a worn 1987 VHS slipcase for an anime video release, seen flat."2. Name its parts: magenta spine band on the left edge, title panel in the top quarter, runtime box at the bottom.3. Cut "8K, masterpiece"; anchor on late-1980s airbrush cover art.
차이를 직접 확인하세요
From test renders to a release schedule in four weeks
사용 전
Orbit Games, a 4-person indie studio. Our launch art looks like everyone's AI art: centered heroes, glow everywhere, gibberish text. We have 12 character reveals and a crowdfunding page to fill before December, and no art director.
사용 후
Keeper rate: 11 percent in week 1, 31 percent in week 4. Week 1: 9 bare prompts audited, average 11/30; Orbit Founders series sheet v1.0 written on F13 trading cards. Week 2: cards 01 to 03 approved in 1.3 rounds on average. Week 3: the Bot's Friday review found keepers for 3 of 4 briefs. Week 4: the dot caught card 04's number as "No. 4/12" the evening before release. Average keeper score: 26.4/30.
One near miss, from Friday review to keeper
사용 전
Fri Oct 16, 2:00 PM, Bot Friday render review: B023, an F05 magazine cover for the Orbit studio newsletter. Best render B023-r1-2.png, 19/30, Needs a pass. Cover lines came out as gibberish and her hair covers the masthead. You answer "log only" and open the Skill in ChatGPT the next morning.
사용 후
Skill, ChatGPT, Sat Oct 17: you paste the Bot's AUDIT and the render. "REVISE for Midjourney." REVISION: B023-r1-2.png, round 2 (Midjourney, seed 5530) Changes this round (3 or fewer): 1. Type zone: "three cover lines down the left side" to "three stacked blocks of small headline text down the left side, too small to read", because the lines rendered as invented words. The real lines are set in layout. 2. Subject: "loose hair" to "hair tied back, the top of her head just below the masthead band", because the masthead is hidden. 3. Type zone: add "a plain blank band across the top 15 percent for the masthead", because the model drew its own title there. Next prompt: front cover of a 1990s monthly science magazine, seen flat, a plain blank band across the top 15 percent for the masthead, a salvage engineer holding a cracked helmet visor, hair tied back, chest up, the top of her head just below the band, three stacked blocks of small headline text down the left side, too small to read, a small blank corner box bottom right, one soft key light from the upper left, slight paper sheen --ar 4:5 --style raw --no logo, rim light --seed 5530 Render B023-r2-1.png. You set the masthead "SALVAGE" and three real cover lines in your layout tool. Bot, Fri Oct 23, 2:00 PM: B023-r2-1.png 25/30, Finished. Hierarchy 2 to 4, Restraint 3 to 4. Listed under Keepers; you answer "approve all", the log gets its rows and the file moves to Keepers. Bot, Mon Nov 2, monthly tune-up: 7 of the 10 F05 renders that were not kept had gibberish cover lines. You approve the new Prompt library rule: for F05 in Midjourney, cover lines are stacked blocks of small headline text, and the real lines are set in layout.
A 12-card series with all three on duty
사용 전
Orbit Founders: 12 character cards, one every Tuesday at 9:00 AM from Tue Oct 13 to Tue Dec 29. Dana owns the series sheet. Series renders go to Series/Orbit Founders in Google Drive; one-off posts go through the Bot's Notion briefs.
사용 후
Mon Oct 5, Skill: SERIES SHEET Orbit Founders v1.0 on F13 at 3:4, seed 2207, 10 drift checks. Pasted into Notion and the Google Doc with the same first line. Thu Oct 8, Dot first check-in: cards 01 to 03 Consistent, 25 to 27/30. Prompt docs for 04 and 05 written. Mon Oct 19, Bot format board: B026, the crowdfunding header, on F24 character select screen, with 8 of 12 tiles dark until their reveal. Thu Oct 29, Dot: card-05-vex-3.png Fix in revision, the accent on two elements and facing left; Final/card-04-tomas-final.png reads "No. 4/12". Both fixed before Monday's release check. Thu Nov 5, Skill: sheet v1.1, "smoke" excluded and one landmark per background. Dot re-checked 06 to 12 and rewrote prompt-07-kit; no approved card needed a rerender. Mon Nov 16, 5:00 PM, Dot release check: No. 06 not approved and no final file. You picked card-06-ida-4.png at 27/30 and exported it by 8:00 PM. Series so far: 6 of 12 released on schedule, 0 Off-format items released, Drift findings per render down from 1.8 on cards 01 to 03 to 0.4 on cards 04 to 06.
구성
- SkillImage Director SkillFormat-first prompts that turn AI images into posters, covers and cards, not test renders.파일 8개 · 1,075줄
- BotImage Director BotA Grok teammate that turns briefs into format-first prompts and scores your renders.파일 1개 · 288줄
- DotImage Director DotYour ChatGPT dot keeps an image series on format, from the first item to the last.파일 1개 · 192줄
여기에 Bundle에만 있는 플레이북이 더해집니다. 파일은 영어로 되어 있습니다.
- Image Director Kit playbook
This is the part only the bundle gets: how the Skill, the Bot and the Dot run as one image direction practice instead of three separate tools.
- What this system does
- Setup order
- How the pieces hand off
- Weekly rhythm
- Metrics that matter
- First 30 days
- FAQ
Kit 소개
Make AI images that look like finished objects instead of test renders: a format chosen on purpose for every image, a score that says what is missing before you reroll, and every series kept on format from the first item to the last. The Skill directs images in the chat where you already write prompts, the Grok Bot plans and scores the week's images on a schedule, and your ChatGPT dot keeps a running series consistent. All three share one format library, one rubric and one series sheet, so a render the Bot scores is revised by the Skill and checked by the Dot without translation. Setup takes about 30 minutes to install the three, plus about 1 hour on day 1 for your first series sheet.
What you get
- Image Director Skill for Claude, ChatGPT, Codex and Cursor. Five modes (AUDIT, FORMAT PICK, FORMAT PROMPT, REVISE, SERIES), a library of 24 visual formats with ratio, parts, type zone and wear, a 6-dimension rubric out of 30, 15 AI-look tells with their fixes, prompt dialects for seven image tools and a worked example from 9/30 to 26/30.
- Image Director Bot for Grok Bot. Five skills and three routines: a Monday format board that turns your Notion briefs into format options and prompts, a Friday review that scores every render in your Google Drive Inbox, and a monthly tune-up that shows which formats and prompts actually produce keepers.
- Image Director Dot for your ChatGPT dot. An ongoing series responsibility: it checks every new render in your series folder against the series sheet, keeps the next prompts written and checks each item the evening before it goes out.
- The Image Director Kit playbook, only in the bundle. The setup order (rubric and series sheet first), how a render moves from brief to keeper across the three, a weekly rhythm that takes about 30 minutes of your time, the four metrics to track, a week-by-week first 30 days and an FAQ.
- One shared contract. The same format IDs, prompt template, AUDIT format and severity scale in every piece, one DECISION format for the Bot and Dot, plus a version line on the series sheet so you can see when any copy goes stale.
Who it is for
- Creators, indie studios and small marketing teams who publish AI images every week and want them to look designed.
- Teams running a series, such as character cards, episode covers or a poster set, that must stay consistent until the last item.
- Anyone tired of rerolling the same prompt and hoping, who wants to know what changed between a 14 and a 26.
What it will not do
- Generate images. You render in your own image tool, under its terms; the kit directs, scores and keeps the record.
- Approve, move, publish or post anything without your answer. Every change to your files is a DECISION you make.
- Copy real brands, living artists, real documents or real people. Formats are described by their structure, issuers are fictional and real people need their consent.
- Replace an illustrator or art director for your signature work. It gets everyday images to finished, and tells you when one needs a person.
Bought separately
The Skill, the Bot and the Dot are $19 each, $57 for all three. The bundle is $39, saves you $18, and adds the Image Director Kit playbook, which only the bundle includes. The playbook is what turns three tools into one practice: the order to set them up in, how a near miss moves from the Bot's Friday review to the Skill for a revision and back for a score, and the numbers that show whether your keeper rate is improving.
리뷰
아직 리뷰가 없어요.
Bundle 더 보기

AEO Kit
Become the brand ChatGPT, Perplexity and Google name when buyers ask who to choose.

Bot Generator Kit
Turn a job you do by hand into a safe, tested AI teammate in under an hour.

Content Factory Kit
One long piece in, a week of on-voice posts out, scheduled in Notion and reviewed weekly.