Post

How to get URL link on X (Twitter) App

On the Twitter thread, click on or icon on the bottom
Click again on or Share Via icon
Click on Copy Link to Tweet
Paste it above and click "Unroll Thread"!
More info at Twitter Help

el.cine

@EHuanglu

May 18, 2023 • 8 tweets • 6 min read • Read on X

Scrolly

twitter.com/i/web/status/1…

Struggling to create consistent models in #midjourney?

I have discovered the ultimate hack.

Let's unlock the secrets of Consistent model-making together!

#AIart #AIArtwork #AIArtCommuity twitter.com/i/web/status/1…

Step 1: Create your favourite model

What’s the trick?

- Clear model face
- Non-stylish dressing
- Clean background
- Use keyword [Minimalism]

Reroll the same prompt until you get your favourite model

Sample Prompt: portrait shoot of a lady, in the style of minimalism --s 1000

Step 2: Find the Seed of your model

How to find?

- Add Reaction
- Type [Envelope]
- You’ll get a notification
- Copy your seed
- Use the seed suffix --seed [Value]

Step 3: Use your model image link

Where to get the link and how to use?

- Upscale your model image, right click and copy the link
- Paste the link at the very beginning of your Prompt

Step 4: Set your style to 50 and use image weight

Prompt suffix: —s 50 —iw 1.5

Why?

- Set style to 50 so Midjourney won’t be so opinionated, you get what you asked
- Image weight is to set your image to higher level

Step 4: Let’s stylise your model

What’s the tricks?

- model image link
- Use same prompt from step 1 but add style keyword
- Use your model seed

Prompt: [Link] inspired by [Artist name], portrait shoot of a lady, sitting on a couch, in the style of minimalism —s 50

Bonus.

https://twitter.com/ehuanglu/status/1658800766033993728?s=46&t=6baTmHNr3371LXRR687Lig

If you are not sure what keywords to use, click here

https://twitter.com/ehuanglu/status/1658800766033993728?s=46&t=6baTmHNr3371LXRR687Lig

https://twitter.com/ehuanglu/status/1658444864105910274?s=46&t=6baTmHNr3371LXRR687Lig

• • •

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

This Thread may be Removed Anytime!

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

More from @EHuanglu

el.cine

@EHuanglu

Sep 13

oh wow.. Higgsfield's new AI Fashion Factory can generate 12 camera angles in one shot.. keep character, outfit and environment consistent

we don't need iPhone 17's pro camera anymore

step by step tutorial:

step 1: Fashion Factory

-> go to Image
-> choose Fashion Factory or here's the direct link: higgsfield.ai/image/soul?mod…

step 2: set style

choose one of the presets, this will define the style of your photos, you can also preview the style

Read 15 tweets

el.cine

@EHuanglu

Sep 11

it's over for.. ad agencies

you alone now can run an AI studio with nano banana + veo 3 on Higgsfield.. any product, effect, talking actor with realistic motion

Veo 3 is 75% off for a limited time now

step by step tutorial:

now.. we shouldn’t be satisfied just by some random clips AI throws at us anymore

what really matters is having full control, actor’s style, expressions, the move, even how they hold the product and the environment around

this tutorial shows how to make that happen

step 1: image generation

Soul + Soul ID + Visual Style - it’s what gives me the most realistic results

I use Soul ID to keep my character consistent across different shots

and for realism, always the Amalfi Summer Visual Style

link: higgsfield.ai/image/soul

Read 14 tweets

el.cine

@EHuanglu

Sep 10

AI is nothing without control..

this new Banana Placement by Higgsfield can add multiple products into any film scene.. use any actor and make them talk, outfit, product, and even add vfx

step by step tutorial:

1. this Banana Placement is powered by nano banana but easier to use

add, remove, or change objects, text, clothes.. everything is editable

you can use brush to select specific parts to edit.. so it's much more accurate

check here: higgsfield.ai/edit?model=ban…

2. go to Edit -> Banana Placement

Read 14 tweets

el.cine

@EHuanglu

Sep 9

wow.. Freepik brings nano banana to next level..

we can use sketches to control actor's pose accurately and then use keyframe to make her dance.. change bg, direct actor and camera..

this is a great way to do film composition

step by step tutorial:

this Visual Prompt feature is very useful for film scene compositing

it can precisely control the pose of any actor, let actor hold your product, change clothes, turn to video, all together

check it out here: freepik.com/pikaso/image-e…

step 1: upload actor image

go to AI Suite -> Image Editor -> upload image

Read 14 tweets

el.cine

@EHuanglu

Sep 2

Apple’s AI getting crazy and nobody’s talking about it

they just open sourced FastVLM + MobileCLIP2, can do realtime VIDEO captioning with phone camera.. runs 100% local on your phone, 85x faster, 3.4x smaller…

free to test and download, link in comment

let's break down:

here’s the problem with AI models, if you bump up the image resolution, the model gets smarter, more pixels, more detail

but the downside is, it gets slower, ’cause higher-res images take longer to process, and you end up with more visual tokens that slow down the LLM setup

some tests here, keeping same data, recipe, LLM but swap out the vision encoder

they compared with ViT‑L/14 and SigLIP‑SO400, a fully convolutional ConvNeXT, and hybrid FastViT models

FastViT is like 8× smaller and 20× faster than ViT‑L/14 while staying just as smart

Read 10 tweets

el.cine

@EHuanglu

Aug 27

this is nightmare.. for actors

this new AI can change what actors say.. to literally anything with their own voice.. no way you’d know it’s fake, this is sooo scary...

step by step tutorial:

step 1: it's called lipsync-2-pro

just drag video and audio on timeline, slide audio around and choose which part of video you want to sync with, this makes @synclabs_so much more efficient

you can also generate and clone voice.. 🔽

link: sync.so

step 2: preparing actor voice

for best voice cloning results:

- use high-quality audio with minimal bg noise
- voice with consistent tone and pace
- at least 1–5 mins of continuous speech
- it can extract directly from video or uploaded audio