Photoshop is DEAD. Long live to FLUX.1 Kontext.

Flux Kontext is an absolute revolution in image editing and compositing.

You don’t need ANYTHING else than a prompt to edit your images.

Here is how to use Flux.1 Kontext

Make your prompt unambiguous

When you want something to change in your image, describe it precisely.
When prompting it’s important to avoid lazy prompts like “make it nice”

• Describe a clear action (e.g. "maintain" or "replace")

• Describe the subject precisely (e.g. "Change the woman hair from blue to dark”)

For instance, when prompting a color change, describe the exact color, not just blonde or blue.

Alright, here are a few images edited with Pykaso AI image editor by using Flux.1 Kontext. 👇

Great, now let’s try changing her location from a simple detailed prompt in:

Edit in iterations

The most powerful feature in Flux.1 Kontext is the capacity to make image edits in steps.
That means, you can edit an image one change at a time.

It’s better to sequence your editing into multiple smaller edits rather than one big prompt because changing many details at the same time leads to more errors in the AI editing.

So here is how you should use Flux.1 Kontext to edit your image in small iterations

Let’s say you want to change the hair color AND the background.

1. Turn a big edit into several “iterations”

2. Start editing the most simple details (e.g. hair color)

3. Then you edit more complex components of your compositing (e.g. background)

💡 You can iterate your edit on the same picture inside Pykaso by clicking ‘Iterate’. This button will take the result from your first edit and allow you to make a new edit.

Let’s say, we want to change the woman jacket color to blue. Just prompt it.

Do you prefer the Eiffel Tower at night? There you go.

Quotation marks

A new prompting concept that Flux Kontext introduces are the quotation marks.
It’s similar to prompting Stable Diffusion models with parenthesis to make a word have more importance.

Here with Flux Kontext you need to add quotation marks to increase the AI model focus on this specific detail change.

For example:
💡 Replace ‘x’ with ‘y’

Character consistency with Flux.1 Kontext

It may sound unreal but FLUX.1 Kontext can generate a completely new image from your input while keeping full character consistency and WITHOUT training.
Let me show you an example:

Prompt camera controls

So character consistency means that from the same source image you can generate different camera angles and framing.
That’s an amazing feature if you want to train a LoRa from a single AI generated image for instance.

Ok let’s try to get a close up portrait from our source image. It’s so consistent 😭

If you want more inspiration here is a table of keywords to test in your prompts with different camera angles and framings

Use cases for image editing with FLUX.1 Kontext

• Remove watermarks from images

• Add objects

• Edit specific areas

• Make a character realistic

• Character consistency

• Professional headshots

• Remove text from image

• Change location

• Combine photos

• Change colors

How to avoid unwanted edits with Flux.1 Kontext?

Flux.1 Kontext is so amazing that usually it won’t change anything except what you asked for.
But if you have unwanted edits or modifications, then all you have to do is write precisely what you want to keep the same directly in the prompt.

You may have noticed that already from my prompts above.

You can do the same with anything you want to keep from the original image:

Here are some keywords you can use to describe something you want to keep

• Keep

• Maintain

• Conserve

• Preserve

• Protect

Conclusion

Flux.1 Kontext is the biggest leap forward this month by far. We AI image editing, compositing and character consistency all in one simple prompting tool.

The best way to learn about it is to start creating now. Try it inside Pykaso now!