Empezar
Reel Pre-Critique: Find Drop-Off Before You Post

Reel Pre-Critique: Find Drop-Off Before You Post

Get a brutally honest critique before you publish. I actually watch your video and tell you the exact second viewers drop off, why the first 3 seconds are weak, and which shots or captions look amateur - section by section, with timestamps. No compliments, only what to fix and how to fix it. Works for Reels, Shorts, TikToks and ad videos.
#Vídeo
Valoración
Se necesitan más valoraciones
Vendidos
0
Cómo se usa
Descargar

The problem this solves

Your own video always looks fine to you. So you post it, it stops at 200 views, and you never find out what went wrong.

This skill lends you an outside eye before you publish. It actually watches the video — the picture and the sound, not the thumbnail or the description — and points to the seconds where people leave.

When to use it

  • You made a Reel or a Short and you are hesitating to post it
  • You posted, the views did not come, and you want to know why
  • You are about to put money behind an ad video and want it checked first
  • You have two or three edits and need to pick one

How to use it

python scripts/critique.py my_reel.mp4
python scripts/critique.py a.mp4 b.mp4          # compare two cuts
python scripts/critique.py a.mp4 --goal "get lesson enquiries"

--goal is what you want that video to get you. Give it, and the standard changes from "is this fun" to "does this achieve that goal".

What the output looks like

This is a real result, not a mock-up. A product intro video, with the goal set to "make the viewer want to hire us to build something like this".

1. Drop-off points
   ...

2. Fatal problems
   Audio, 0:00 to the end. A flat, slow machine voice drags the
   credibility of the whole video to the floor.
   Picture and pacing at 0:08. It claims a 30-minute job now takes
   one second, but the screen is static, not one click, no visual payoff.
   Captions at 0:06. Large clumsy captions in the middle of the frame
   cover the very screen that proves the program works.

3. The real reason it misses the goal
   The video does not demonstrate the skill or the expertise that would
   justify paying for this. The video itself looks cheap, which leaves
   the impression that whatever was built is cheap too.

4. If you cut it again
   Cut the dead air and raise the pace of both audio and picture.
   Replace the machine voice with a human one you would trust.
   Stop holding the whole screen still — zoom into the part of the UI
   where the result appears.

5. Post it as it is?
   No.

What you need

  • One Gemini API key (environment variable GEMINI_API_KEY, or .env)
  • pip install google-genai

Cost

It scales with length. A 30-second Reel is roughly one US cent, a one-minute video about two. Ten videos a day is still small change.

Design principles

No compliments. Ask "please evaluate this" and you get nice words first. So the critic is seated as a judge and cannot write a single compliment.

Timestamps required. A note you cannot act on is not a note.

Says "I don't know" when it doesn't. An invented criticism is worse than no criticism.

FAQ

Which video formats work? Ordinary video files such as mp4 and mov. Vertical (Reels, Shorts) and horizontal both work.

Does it handle long videos? Yes. Cost scales with length, so check the cost first for anything over ten minutes.

Does my video leave my machine? It is sent to the Gemini API for analysis. Do not put confidential footage through it.


API externas

generativelanguage.googleapis.com