TechHalla Profile picture
Apr 18, 2023 6 tweets 3 min read Read on X
(1/6)🧐Character Consistency for Video Games in #Midjourney

🧍‍♂️Introducing Jimmy Jazz, a potential character for a Lucas Arts-style 2D adventure.

🧵In this thread, I'll show you how to do it so that you can apply it to your own projects.

#AI #AIart #conceptart #videogames /imagine prompt: a comic dr...
(2/6)The tricky part is keeping our character consistent while applying variations.

So let's do just that. Copy the URL of the initial image (ensure to open it on browser first), use /imagine prompt:, and then paste the same prompt as before. But let's add a little modification /imagine prompt: [url_of_up...
(3/6) Now, do the same, maintaining the original image url, but modifying his status. Lets say... bowing! /imagine prompt: [url_of_up...
(4/6) And keep going on! Now Jimmy is prepared to run /imagine prompt: [url_of_up...
(5/6) We can also add accesories, such as sunglasses /imagine prompt: [url_of_up...
(6/6) I know the results can be improved, but certainly this can greatly accelerate the process of designing characters, animations... Give it a try and let me know!😊

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with TechHalla

TechHalla Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @techhalla

Aug 12
looks just like a real video game, right?

well, it was made 100% with AI, and I’m gonna show you how you can make one too in this thread 🧵👇

includes prompts and tricks, so bookmark it for later
Let’s start by creating the character.

We’re gonna do everything from @LTXStudio, and I’ll drop the access link at the end of the thread. Go to image generation, pick a 1:1 ratio, and use the prompt you see in the image.

Tweak it to fit your needs! videogame 3D character. Back view of a fat 3D CGI character rendered in Unreal Engine | man standing with relaxed posture | wearing a faded blue denim jacket with shearling collar, slim-fit black pants, and black boots | slicked-back hair with visible individual strands | clean white studio background, even lighting, soft shadows | ultra-realistic textures, high-detail fabric and material rendering, 4K render
Image
Now let’s make the first scene. Upload the character image, set the aspect ratio to 16:9, and have it build the scene around it with relight and scaling.

Prompts are in the ALTs, and you can tweak outfits or props too.

Upscaling the results is optional but recommended. keep the character from behind and include a realistic 3D American city sidewalk rendered in Unreal Engine | cracked concrete pavement with faded yellow curb markings | street lamps, trash cans, and scattered paper debris | brick building façades with shop windows and awnings in the background | soft afternoon sunlight casting long shadows | ultra-realistic materials and lighting, high-detail textures, AAA video game environment  in the background
keep the character from behind and include Realistic 3D interior of a clothing store rendered in Unreal Engine | dim warm lighting with soft shadows | stacks of folded clothes on wooden tables, hanging garments in various colors on racks | textured fabric materials with visible stitching and folds | at the back, a wooden counter with an elderly male shopkeeper standing behind it | subtle reflections on polished floor | immersive AAA video game environment with ultra-detailed props and lighting in the background
Image
Image
Read 8 tweets
Aug 10
is it possible to make videos longer than 8s with Veo 3?

with LTX Studio and this super simple workflow, you sure can.

thread with all the info, prompts included 🧵👇
1. Let’s start with the base image.

I used Flux Kontext in @LTXStudio and the prompt you’ll find in the image’s ALT text, using Vintage as the reference style. Image
Third-person POV from behind of a cyberpunk synthwave woman wearing retro 80s attire: metallic cropped jacket, high-waisted neon leggings, fingerless gloves, and chunky sneakers, walking down a neon-lit cyberpunk city street under soft daylight | realistic photography | ultra-detailed fabrics and skin textures, cinematic lens, shallow depth of field, lens flares on neon signs, light atmospheric haze
2. Make a video with Google Veo 3 Fast without leaving LTX, using the previous image as a reference and the prompt you can find in the ALT.

Grab the last frame of that clip. We’ll use it right after. Image
{ "scene_description": "Cinematic camera following a woman walking down an urban street lit by neon signs at dusk.", "visual_style": "Cinematic, realistic, with a nighttime urban neon aesthetic.", "camera_movement": "Smooth tracking shot from behind, following the woman as she walks.", "main_subject": "Woman wearing a silver jacket and colorful leggings walking down the street.", "background_setting": "Urban street with skyscrapers and glowing neon signs, people walking along the sidewalks.",...

Image
Read 6 tweets
Aug 9
5 minute guide to creating Midjourney video compositions with depth maps in DaVinci Resolve

Includes prompts and tips 🧵👇
Step 1

Create the main image for your video along with any extra elements you want to add in.

I used Midjourney and the prompts you’ll find in the ALTs. Wide shot from a slightly elevated angle, full side profile of a black woman with short black hair, sunglasses, wearing large over-ear headphones, dressed in loose oversized streetwear, lying on a rooftop ledge positioned to one side of the frame | pronounced cinematic depth of field with sharp foreground and blurred distant city skyline at sunset with dreamy bokeh lights | muted warm and teal color palette | subtle vintage film grain | lo-fi chill mood | soft golden-hour ambient light --ar 3:2 --raw --stylize 240
A single colorful hot air balloon floating in the distance against a completely flat pure white background | minimalist composition | negative space emphasis | cinematic framing | ultra-clean high contrast aesthetic --ar 3:2 --raw --stylize 240
Distant silhouette of a seagull, captured mid-air against a completely flat pure white background | composition | negative space emphasis | cinematic framing | ultra-clean high contrast aesthetic --ar 3:2 --raw --stylize 240
Step 2

Animate the images (in SD so you don’t waste resources). I used auto mode with low movement.

For the main one, I extended it to get the longest clip possible, using the first frame as the last one so it loops perfectly.

Read 7 tweets
Aug 5
Tired of making pixelated, low quality videos?

Higgsfield just added an AI video upscaler that turns your creations into Hollywood-level stuff, like shot with a $200K cinema camera!

8 crazy examples and everything you need to know 🧵👇

1/8
2/8 With Topaz now part of their suite, @higgsfield_ai got everything you need for your videos.

Create eye-catching visuals, add their effects, and upscale them for even more impact.
3/8 Doesn't matter if your videos are packed with motion or pretty chill...

This video upscaler can handle it all! It's the new standard for cinematic visuals on an AI platform.

Check out the details 👀
Read 12 tweets
Aug 3
you just need one AI tool to make these videos with consistent characters and style.

I'll reveal all the prompts, tricks, and secrets in this LTX Studio thread tutorial 🧵👇

It even includes prompts for Flux Kontext and Veo 3.
Start with the story: plan shots and create required stills.

Do this first, then assemble videos for a well-organized narrative.

I used @LTXStudio image generator for the initial image and derived new stills from it. Now, I'll explain.

Prompt in the ALT. A rugged 40-year-old man with a beard and a cap, wearing denim clothes, holding fast-food bags labeled "BQ" with a crown logo | casual medium-shot photograph inside his home | blurred living room in the background | natural lighting, candid style
Using that image as a reference, I asked it to generate another shot, via natural language prompts.

From there, with more prompts, I edited by removing and updating elements until I got the exact frame I wanted.

All details in the ALTs. swap the camera to the opposite angle. showing a woman in her 20's putting dishes on the table. dressed in casual 70s outfit, she has black hair.
remove the man on the photo. keep the rest intact
remove all the dishes
the woman with 70s pageboy hair, is holding a white dish with his left hand. put a tablecloth over the table. 70s aesthetics
Read 7 tweets
Aug 2
A lot of people have been asking me how I made this AI video, with consistent characters, styles, and realistic lip sync.

I’m breaking it all down for you in this thread 🧵
Let’s start with the character.

I used Midjourney and the prompt you can see in the ALT text of the first image.

Here are the other 3 outputs too. A gritty portrait of a rugged, heavily tattooed bearded metalcore singer with piercings, scars, messy dark hair and a worn black leather jacket, staring intensely at the camera with a serious, intimidating expression, standing in a decayed abandoned lot with broken walls, graffiti and scattered debris | dramatic cinematic lighting with harsh shadows and warm rim light | slightly blurred background for depth | raw, rebellious metal atmosphere | professional concert photography style
Image
Image
Image
Still in Midjourney, I animated it. I used automatic animation with low motion.

We just need the camera to move a little and the character to feel alive.

Don’t worry about the lip sync, we’ll take care of that next.
Read 8 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(