← All transcripts

This One Prompt Fixes AI Character Consistency (Use This) Transcript, AI Summary & Key Points

AI Samson · Feb 27, 2026 · Science & Technology · 13:43 · EN

Watch on YouTube

AI Summary

A repeatable character-reference prompt creates a sheet showing a character from multiple angles, including full-body front, side, three-quarter and back views plus facial close-ups. The reference sheet can maintain character identity across images and videos while preserving difficult details such as scars, tattoos and piercings. Google Flow with Nano Banana supports image references, image-to-video workflows, lighting and outfit changes, multi-character scenes, and age transformations. Creating an image first and using it as the video's first frame provides more control than directly turning an image reference into video. Consistent character references also support digital personas and AI-generated content at scale.

Key Points

  • A single well-defined prompt generates a character sheet with front, side, three-quarter and back full-body views, plus front and profile face close-ups.
  • The method works across image models and platforms; it was tested in Nano Banana 2 and demonstrated mainly with Nano Banana Pro in Gemini and Google Flow.
  • Midjourney provides an image reference for starting a character, while the character-sheet prompt specifies poses, layout and a hyperrealistic rather than 3D-rendered style.
  • Four outputs provide useful variation because the correct result rarely appears on the first AI-image generation.
  • Google Flow can upscale results to 2K, and 4K upscaling is available on the highest tier.
  • Full-body views may omit parts of the character, but adding an explicit instruction that the entire top row must show head-to-toe views with feet visible and no cropping can fix the issue.
  • The character-sheet approach maintains distinctive details across instances, including a strong facial scar, complex face tattoos and specific piercings.
  • For direct image-to-video generation, an image reference can be added as an ingredient and referenced in a prompt such as creating a video of a woman commanding a spaceship.

AI in practice

Used for

Business ideas

Create a consistent AI character, produce images and videos with that character at scale, and use the resulting content to build an audience and pursue online monetization.

For
People who want to create online content without showing their face, being famous, having a team, or relying on a real person to portray the character.
Solves
AI-generated characters can change appearance between images, scenes, and videos, making it difficult to produce a recognizable persona or consistent film character.
  • A female protagonist was used in a character sheet, a spaceship-commanding video, a futuristic-motorbike scene, outfit changes, and facial-tattoo tests; the transcript claims the character remained consistent.
🔒  Build steps and tools for 1 idea. Unlock

Tools & resources

5 items

CNo. 0016
AIAINotes.us AI product

ChatGPT

chatgpt.com

ChatGPT is a general-purpose AI assistant that can generate movie ideas and expand them into story and storyboard prompts. It can also produce software projects from natural-language prompts, such as a Python dating application with user profiles and swiping logic, and help explain electronics and hardware-development concepts. It is also cited as a general-purpose AI capability that consumer products can package into more personal, character-based experiences.

Mentioned in
81 videos
Kind
AI
GNo. 0098
AIAINotes.us AI product

Gemini

gemini.google.com

Gemini is Google's AI model and assistant accessible through gemini.google.com. It processes mixed media including video, audio, images, and text, and can convert recordings into SOPs, summarize meetings, extract decisions, and draft follow-up emails. In research and development contexts, it serves as an alternative LLM backend for tasks such as adversarial audits of requirements and resume evaluation.

Mentioned in
27 videos
Kind
AI
GNo. 3344
AIAINotes.us AI product

Google Flow

labs.google

Google Flow is a visual-creation platform for generating and upscaling character sheets, creating reference-based images, and producing image-to-video clips. Its workflow uses a character reference sheet that can be reused across image and video generations to maintain a character's facial identity, including across scenes with multiple characters, different outfits or lighting, scars, tattoos, and age changes.

Mentioned in
1 video
Kind
AI
MNo. 0076
AIAINotes.us AI product

Midjourney

midjourney.com

Midjourney is an independent research lab and provider of a text-to-image generation service developed by Midjourney, Inc. Its image-generation tools—accessible primarily via a Discord bot and a web application—create images from user text prompts using proprietary AI models; the service launched publicly in 2022.

Mentioned in
5 videos
Kind
AI
NNo. 0186
AIAINotes.us AI product

Nano Banana Pro

In the AINotes directory

An AI image-generation and image-editing model from Google, also known as Gemini 3 Pro Image. It generates images from text prompts and can create or modify specific visual compositions, such as depicting a model holding a supplied object.

Mentioned in
2 videos
Kind
AI

Links mentioned

🔒 Full analysis locked

Unlock more videos and the full analysis

A credit unlocks one video's full analysis for good — the build steps, the tools and how each was used, the methods behind every use case. Pro opens the whole library instead, and raises how many videos you can analyse a day.

Unlock full analysis — free

Transcript

Searchable transcript of This One Prompt Fixes AI Character Consistency (Use This) — AI Samson (13:43). Search for a phrase, then click its timestamp to jump straight to that moment in the video.

Captions sourced from the original video on YouTube, published by AI Samson. The video, its captions and all related intellectual property remain the property of their respective owners; AINotes claims no ownership. Provided for research, accessibility and search — see the Transcript Notice and Copyright Policy.

00:00 This one prompt just changed the way that I create consistent characters with AI forever. I stress tested this across a number of complex use cases like facial tattoos and strange scars as well as very specific piercings. And the results were truly incredible. Yet nearly everyone is still over complicating the character consistency process. So, I'm going to show you how you can lock your character consistency once and for all using one simple prompt.

00:30 Let's take a closer look at exactly how this works. Now, using one simple well- definfined prompt, we can get out this perfect character sheet which displays to us our character from each angle. We get the whole body shown from the front, from the side, from a three/arter view, and from the back. We also get the face close up from the front and from the profile.

00:53 Now, this gives us exquisite references to then go out and make videos just like this. So, let's do it together. If you want to follow along with this, I'll leave a link to all the problems that I'm using for free in the description below. This method works across all image models and platforms. The power is in the repeatable system. I've already tested it inside Nanobanana 2, and it's even stronger.

01:31 For most of this video, you'll see me using what was called Nano Banana Pro. And it's possible for you to use this either inside of Gemini or where I prefer to use it, which is in Google Flow. Now, I prefer flow because it's specifically designed for visual creation. Now, we have two options here. You can even start creating your own original character inside of Flow or you can take a reference from somewhere else.

01:58 I find the following image model to have the highest degree of aesthetics and that is our old favorite midjourney. So, to show you this, I will be taking a recent exploration that I created and I'm going to be working with this woman here. Now the next step is to upload it as a image reference. Now you can go ahead and copy the prompt from the first page of the prompt guide and pop that into your prompt bar.

02:22 Now to give you a quick summary of what this prompt entails without reading the entire thing. It asks us to give a very specific layout where we are seeing our protagonist in some very specific poses. It also outlines exactly what style of reference sheet that we want. that makes sure that it's hyperrealistic and it doesn't fall into creating like a 3D render style.

02:46 Now, you can select the number of outputs. I find four is the best approach. And the reason for that is one of the key caveats with working with AI imagery is that it's very rare that you get the right output on the first take. Now, what's incredibly remarkable about this is just how visually intelligent Nano Banana is. Let's take a look at what I mean.

03:07 We have this perfect rendering of all of the elements that we are looking for. It's exquisitly captured all of the detail. Now, what you can do, of course, is upscale this directly inside of Flow. And to upscale it, all you have to do is go to download and then select 2K. You can even get 4K now upscaled if you're on the highest tier. But it's interesting to look at the different variations we get out.

03:30 And as you can see, each one has a slightly different interpretation of what we create. Now, you'll also notice that sometimes it doesn't always accurately give us the head to toe feature. That's one of the slight bugs of this, but if you go through a couple of iterations, it will do. So, the other fix to this is to take the one that you like and then explicitly add that as an image prompt and then say remake this image, ensuring the entire top row shows full body views from head to toe.

03:58 All subject in top row must be fully visible including feet with no cropping at the ankles, knees or head. As you can see, once we add this prompt and edit it, we get an updated version with our feet included. Now, to show you a few examples of just how powerful this is, especially with some complex edge cases, take a look at these. Here we have a animated character and you can see from the output that it's possible to get a perfectly rendered character sheet of this little crocodile chap as well who's a little cutie.

04:31 How can we take this even further? Well, we can look at adding extremely unique artifacts to our characters which are harder to recreate. For example, scars and tattoos. So, as you can see, I've created this gentleman with this very strong scar coming down the side of his face. And you can see that across all instances, it is perfectly consistently rendered.

04:54 Now, we've got these wonderful images, but of course, this is just the start because what we want to start to do is create whole sequences of our character. And there are a couple of approaches that you can use for this. Now, the first method is the quickest. And to do that, what you can do is take your image reference and in Google Flow and use it as an ingredient.

05:15 And to do that, all you have to do is go to the drop down and switch to ingredients to video. Now, other tools do have their own equivalents of ingredients, too, but sometimes they name them a little bit differently just to confuse things. Now, once you've popped the image in, you can reference it in the prompt. So, you can say, "Create a video of this woman commanding a spaceship."

05:35 Then you can select the amount of outputs you want and go ahead and generate that. But what I think is the best approach because it gives us more control is to first create an image and then use that as the first frame for our video creation. That's because we get a bit more control over exactly what the video outputs will be. So to do that, it's extremely simple.

05:58 All you do again is switch to create image, insert our initial character reference as an image reference, and then create an image referencing our character. So I've asked it to show this character riding a futuristic motorbike through a dark neon lit city at night. And I will get four images for this and go ahead and generate. Now we get our output.

06:20 All we have to do is go to frames to video and we can use this as the first frame of our video. And I found that this gives us a much higher quality result. Now the next fun thing we can do is start to combine our characters into scenes. So we might take this young lady and I will also be working with another character I've been developing which is myself as James Bond.

06:42 Then I can add my myself to the prompt and we can define the exact interaction that we would like to see between our two cast members. So, I've asked to show these two characters standing and talking at the edge of a cinematic bridge. Now, you might be asking how I get these very long, detailed, and specific prompts. Well, for that, I use metaprompting.

07:01 I take my initial idea and then ask ChatGBT to specifically enhance my prompt, and consistently it gets me better results. I'll also include that metaprompt inside of my prompting guide. Here we have me and this lady talking. >> You think you can defy me here? >> We need to reconsider. It's too dangerous. >> Failure is not an option. >> But there's so much more we can do.

07:25 And I want to show you some other creative ways that we can use this method. And the first is by adjusting the lighting. So what we can do is we can take our reference sheet and then use the lighting prompt from the character sheet prompt guide. And we can then completely adjust the lighting. This means that we get the ability to move between completely different lighting situations.

07:49 We can take this neon lit situation and use it to reference future videos. But we can do so much more because we can also go ahead and change the outfits of our characters. So I've taken our female protagonist and I've outfitted her with some metallic armor. Now you can also see her in some sort of imperial loungewear. And to do this, it's extremely easy.

08:08 All you have to do is take the reference sheet, use it as an image reference, and then the following prompt. Let's do one together. You I've taken the reference sheet. I popped in the prompt. Now, in the prompt, there is the option to input your own definition of a outfit, but what's also cool is that you can use an image reference as an outfit. So, I'll be using this red cocktail dress.

08:30 And I'll upload that as an image reference. Crop and save it. And go ahead and generate. And here we have our character. And as you can see, she is wearing this lovely red cocktail dress. But let's try and push this as far as we can and see if we can break it. So let's hypothetically say our character in our film wants to go ahead and get a complex face tattoo.

08:54 So we have our facial tattoos and we can go ahead and generate some images and see if we can maintain this perfectly consistent tattoo with such a complex use case. Now there's also another way we can do this which is simply by not using an image reference and using a text prompt to define our character straight away. Now, this cuts out a stage, but I would say that you are losing some control with that approach.

09:19 But there is a prompt in the guide to do it that way, too. And as you can see, you can still get out some extremely useful results. This could be perfect if you're looking to create some marginal characters who only appear periodically. Now, let's take a close-up look at how well it's rendered out this face tattoo. And you can see that it maintains wonderful consistency.

09:38 Now, another extremely impressive way that we can use this method is by aging our characters. And this can be incredibly useful if you're looking to show flashbacks or flash forwards to different eras of a charact's life. So, I thought I would do it on myself and see how I'm going to look in about 15 years time. Looks as though I'm holding up pretty well.

10:00 And to do this, I've also included a prompt which you can simply copy and paste, use the character reference, and then use that prompt. And all you have to do is change the number of years. So this is me by 15 years and here's me by 50 years. We can actually use these to show fantastic time-lapse transformations of characters aging. So here is one example of myself aging from the current age all the way through to 50 years in the future.

10:25 And to do that it's very simple. All we have to do is generate a first portrait-like frame of ourselves or a character at one age and then we can create a second frame of them at a different age and use those two as references in image to video. So all you need to do is go to frames to video, insert the first frame as the young version and insert the second frame as the older version and then you can use a prompt like uh time-lapse between these two ages and you get out this very impressive result.

10:55 Now, one thing that this type of character consistency unlocks is the ability to create effective digital personas. And I think this is going to be one of the hugest trends and biggest opportunities for making money online with AI. And if you are interested in that, I'd love to tell you about today's sponsor. Now, AI content creation is a huge opportunity, but let's talk about how we can actually monetize our AI content.

11:23 Right now, content creation with AI is one of the most interesting emerging business models. You don't need to show your face. You don't need to be famous. You don't need to be a hot girl. You don't even need a team. You can build an AI character, generate content at scale, and grow an audience fast, all with AI. Now, what's great about this is it's early, and when you're in early, the leverage is incredible.

11:46 This is the future of content creation, and that's exactly where Fan View comes in. It's the first platform built specifically for AI first influencers. While other platforms are still debating whether AI creators are allowed, Van View is fully embracing them. You can post, grow, and start earning immediately. It's already the most AI friendly platform out there and it's leading the charge.

12:09 Inside my AI influencer course, I walk you through exactly how you create your own AI character, how to grow it, how to craft it, and how to launch it in the right way. And in this course, I specifically recommend Fan View as the platform to monetize your character. People are already making thousands a month doing this, and the waves only getting bigger, and it's the ones who start now are the ones who are going to win.

12:36 So, be bold and build your own influencer. Check out both links in the description below, and a big thanks to FanView for sponsoring this segment of the video. Creating highly consistent characters is the holy grail for AI film making and now it's much easier with the ability to use effective character reference prompts. We've looked at how you can create your own reference sheets using this highly helpful prompt.

13:03 We've taken a look at ways that we can change the character's outfit, change their age, change the lighting conditions, and also animate different characters together consistently. This unlocks a whole new world of AI film making possibilities. Download my free prompts in the description below. And if you enjoyed this video, then watch this one next, which is all about some powerful cinematic film making techniques.

13:28 And it's a great place to apply this type of character consistency. But most of all, I want to say thank you so much for watching. I truly appreciate you being here. And I wish you a delightful