Chase Lean Profile picture
Mar 5 8 tweets 3 min read Read on X
I think Ideogram 1.0 is a real breakthrough in AI image generation.

Its ability to follow instructions is way better than MJ, and the image quality is superior to DALL-E.

Here are 5 images I generated, that are almost impossible with MJ and DALL-E 3:
Prompt:

A four-part series depicting the life cycle of a butterfly: "Egg" shows a tiny egg on a leaf, "Caterpillar" captures a hungry caterpillar eating leaves, "Chrysalis" illustrates a chrysalis hanging from a twig, and "Butterfly" presents a vibrant butterfly spreading its wings, each stage beautifully showcasing transformation.Image
Prompt:

A photo of a boy holding a blue ball in his right hand, and a green ball in his left. Beside him is a wooden table, with a large pink ball and three smaller yellow cubes. The pink ball should have the number "10" on it. Image
Prompt:

A series of portraits capturing the same woman at different stages of life: The first shows the face of a curious young girl, wide-eyed and innocent. The second portrays a teenager, her features marked by the first signs of maturity and a glimpse of independence. The third image reveals the woman in adulthood, her expression confident, bearing the subtle marks of experience. The final portrait depicts her in old age, her face etched with wisdom, each wrinkle telling a story of a life fully lived.Image
Prompt:

A four-stage illustration showing the evolution of a cannon upgrade in an RPG game: The first image, "Cannon Lv1," depicts a basic, sturdy cannon. The second image, "Cannon Lv2," shows the cannon with enhanced barrels and reinforced metal. The third image, "Cannon Lv3," illustrates the cannon with advanced aiming mechanisms and magical runes. The fourth image, "Cannon Lv4," reveals a highly sophisticated and deadly cannon, glowing with energy and ready for battle, showcasing the progression of power and technology.Image
Prompt:

A cinematic montage showcasing the four elements: "Air" with a breezy sky filled with floating clouds, "Water" depicting a deep, serene ocean, "Earth" capturing a lush, green landscape, and "Fire" illustrating a dynamic, crackling bonfire, each embodying the essence and power of its respective element.Image
Ideogram can really follow long, complicated instructions. All pics above were made in 1-2 tries.

It's even better than DALL-E 3, and unlike it DALL-E, you can easily get it to generate photorealistic images.

I'd say DALL-E is obsolete, until possibly when Sora releases.
As for MJ, it still rules at artistic flavor.

You also have more control over the style, parameters, variations and inpainting, which Ideogram can't.

But if you want to generate decent, realistic images that REALLY follow your prompt, Ideogram is the way.

Plus, it's free.

• • •

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

Keep Current with Chase Lean

Chase Lean 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 @chaseleantj

Mar 7
Here's an easy way to change the style of an image but keep the same character and pose:
Image
Image
Use the image prompt feature in @getremixai.

While there are other options available, this method has a few benefits I'll explain later.
First, upload the image.

Then, set the image weight to Low.

Choose your model and turn on Preserve Composition.

Preserve composition acts like a ControlNet to keep the pose the same.

After that, type your prompt and generate the image.
Image
Image
Read 6 tweets
Feb 3
I just found a way to create both colored and uncolored versions of the SAME image in Midjourney.

This is perfect for making coloring books.

Here's how: Image
First use this prompt:

On the left, a black and white simple children's coloring book illustration featuring a [X] with no shading, with a white background. On the right, the same illustration with color. --ar 2:1 --style raw --sref [see image below] --stylize 250 --v 6
Next, use this image as a style reference.

It makes the prompt much more consistent. Image
Read 5 tweets
Jan 27
Character consistency is a big problem when making AI images.

I’ve been testing a new way to solve character consistency, and it’s by far the best method I’ve tried so far.

Let me show you how: Image
We’ll be using the website @scenario_gg (link below).

Scenario allows you to train custom SD models on your character – and it’s both really easy and completely free.

Having tried it, DAMN it’s good. Image
Before training a model, you first need 5-15 reference images of the same character.

The easiest way to get these images is to make a character sprite sheet with DALL-E.

Super easy. Image
Read 12 tweets
Jan 21
I've generated 10,000+ images with DALL-E 3.

My Logo Creator GPT is also the #1 GPT under DALL-E.

Here are the 10 most useful DALL-E prompting tips I've learnt: Image
1. DALL-E has no negative prompts.

To remove an object (like a gun) from an image:

Use a positive prompt that is the opposite of the unwanted object. Example:

Doesn't work: "No gun"
Works: "Family friendly"
2. DALL-E inside ChatGPT can only generate one image at a time.

But if you want more images without wasting your chats:

Ask it to generate X images, *one after the other*.

Works like magic.
Read 14 tweets
Jan 18
Midjourney and DALL-E 3 can generate some great product photos.

But you can’t place your *actual* product into those images.

I’ve been testing a great tool that solves this problem.

Here’s how to use it: Image
Before we start, here’s a quick demo of a photoshoot I generated.

The perfume bottle is real, but the background pics are made by AI.

Note that it also creates shadows and reflections!

Given that professional photoshoots cost thousands, this is very impressive. Image
The website is called ProductScope (link below).

Start by uploading your product photo.

Then, head over to the “Themes” section, and pick your background scene.

If you like, you can also use a custom prompt to generate 4 high quality scenes and pick the best. Image
Read 6 tweets
Jan 6
Midjourney vs DALL-E 3 (text generation)

A few hours ago, Midjourney dropped a big update to v6 - and it's now even better at making text.

But is it better than DALL-E?

I ran 15 experiments - and here are the results:
1. Prompt: The words "THE LEGEND 27" on a movie poster featuring a legendary sword, and flames in the background

Left: DALL-E 3
Right: Midjourney v6 (updated)
Image
Image
2. Prompt: A man standing alone in a dark empty area, staring at a neon sign that says "EMPTY"

Left: DALL-E 3
Right: Midjourney v6 (updated)
Image
Image
Read 20 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!

:(