Hello,

Since I launched Dezygn 4.0 two months ago, I received tons of feedback from you which help me steer Dezygn in a new direction.

What I heard from most was that you felt like Dezygn was this big sandbox, but you didn't know what to do in it.

Today's update was made possible only by you. Keep the feedback coming and let's make Dezygn the best platforms for freelancers and agencies.

The Dezygn Shortcuts update core goal was to make you reach your visual marketing goals faster. Watch the summary here (with timestamps) or keep scrolling.

Want the full story? Watch the 30 minute video update here

At the heart of this update lie a better system infrastrucutre which systematically creates better designs.

The first major improvement to Dezygn is the quality of dezygn and prompt adherence.

By improving the underlying infrastructure beyond our models, you can now confidently create accurate representations of your models in various situations.

Take this image of this bag here.

Using a single prompt, we can now reference it in various poses without losing any of the bag's essence. This even reveals angles of the bag which it wasn't aware of before.

And here is the same bag being held by our virtual model.

The latest update also brings a stronger adherence to prompt.

One of the struggles with multi-assets image to image (adding multiple ingredients to generate a single image) is that the AI understands what's being asked of it.

In most cases now and with sufficient details, Dezygn can accurately assemble together multiple images in very complex scenarios. We will showcase more of these examples in the Collage update below.

This also applies to our reply-to-image aka editing feature.

You can now get much better results when requesting an edit.

For a lot of our users, text adherence is very important.

And more than often, when the AI returns text which is all mangled it would often ruin your designs.

I am glad to report that the text fidelity has been vastly improved.

It is not perfect yet, but while earlier versions were struggling with more than 2 words of text.

Now this version has gone up to 10 lines of text, which in our testing over 60 words on average, and 20x better than before.

Consider these examples.

In this Dezygn image, on the first attempt, you can see that text fidelity is perfect with less than 4 lines of text.

​10 lines of text

With 11 lines of individual text given to Dezygn, the result is 100% accurate. There is no mistake at all at this stage which is a massive improvement over previous versions which were struggling with more than 3 words.

​20 lines of text

This is our most challenging task for Dezygn. At 20 lines of text, we can see some noticeable text accuracy and layout issues which would not make this visual commercially viable on the first shot.
​
This said, in post-edits I was able to fix some of the wording by targeting individual changes.
​
Generally speaking, Dezygn is not yet ready to produce full commercial brochures and 3 pages long infographics. For the rest, I think we are there.

Here is a 10 lines infographic generated in one-shot by Dezygn:

For the longest time, I scratched my head wondering how to solve the Prompt Paralysis problem.

When you start a new session with Dezygn, unless you are a professional graphic designer, illustrator, photographer and videographer, you wouldn't know what to ask for.

There are so many subtleties and industry-specific knowledge that comes with creating great designs.

For instance, wehn you ask Dezygn to create a logo for you, which of the 6 types of logo do you want? A wordmark, a lettermark, a combination mark, an emblem, a mascot, an abstract mark or a pictorial?
​
And when you want an image of a "woman hodling your product".. Should she hold it with one hand, both hands? Should the camera be at a 45 degrees angle, top-view, close-up or full-body?
​
There are so many parameters which we just can't leave to the AI to guess for you.
​
You need a way to direct it. In fact, with Dezygn, you become a creative director.

This is what I set to solve with the recipes update.

You can now access it in Dezygn by clicking on the little magic wand 🪄 icon in the chat.
​
This opens a series of recipe which makes creating amazing designs an incredible experience.

When you open teh book cover recipe, this is what you see:

Let me break it down:
​
​Ingredients
​
This is where you can optionally add your logo and your cover image.
​
​Shot Composition
​
This is the core innovation here. With the shot composition, you can easily choose from multiple types of covers.
​
But you can also customise this prompt and make it your own on the fly.

This new UI unlocks two things which were missing from virtually all AI platforms so far: guidance and personalisation.
​
You are not just getting "pre-made" prompts which are stiff and hard to work with, and lack options.
​
You are getting dynamic, multi-faceted, and customisable prompts which you can enhance with your own imagery to create works of art which have been unimaginable so far.

Recipes change the game.

I am a big fan of mockups.
​
So much that nearly 10 years ago, I created Mockzign, an amazing mockup generator with over 400 different mockups.
​
But Dezygn now takes it to a whole other level.

Mockzign had 400 templates at the time.
​
Dezygn technically, can support 1000s.

Another recipe here is website mockup. Give it a screenshot, and the AI will add your screenshot into a device screen.
​

Another series of recipes I included are the product in scene, product with model recipes.
​
This is perfect for e-commerce, and coupled with Dezygn's better prompt adherence and text fidelity, makes Dezygn the ultimate solution to power any Shopify / Amazon / Big commerce listing images.

The Product in hand and product with model recipes also deserve an honorable mention.

The recipes coupled with the Dezygn ui improvements introduces something that was not possible before.
​
Instant avatars and models, re-usable in any context, in less than 30 seconds.

To give you some perspective, the previous Lora tech which enabled the training of custom AI avatars took about 10-15 minutes to train a single character (and it did not always work at first, often requiring fine-tuning).
​
It also used to cost about $5 to train an avatar. So it was lengthy and expensive.

Now you can do the same in Dezygn near instantly.

Simply choose the Professional Model recipe in Dezygn.

In a single shot, it generates your model in 9 different poses in a single image. This allows you to then attach it to new prompts, and because your AI has so many references of your model, it ensures consistency.

Combine it with the Product + Model recipe for amazing results.

Now what if you wanted this model to interact with a specific set of props?
​
Like a pair of shoes, a jean, a t-shirt, a pair of glasses and a bag?

I am proud to say that Dezygn now enables that with the collage feature, available in the Ingredients modal.

See that little "Create collage" button?
​
This generated a single image like this.

This means that while you were limited to a max of 4 ingredients in the past, you can now collage together 10-15 ingredients in a single design.

I know it's a lot...
​
But I kept the biggest update for last.

Templates.

Templates come to supplement recipes.
​
So that not only you can upload your images as ingredients to create the design.
​
Not only do you benefit from the Shot Composition engine which makes prompting 100 times easier.

But now you can also benefit from our templates library, starting with Facebook Ads.
​
How it works

I curated 600 Facebook Ads creatives by hand, proven winners from big brands.
​
This will save you $100s of dollars in trial and error, because you can start instantly with amazing creatives.
​
I can't stress how game changing this update is.

Let's see how it works.

First in the recipe list, choose the "Facebook ad" recipe.
​
​

Then, when you click on Choose template you are taken to the Facebook Ad library.

After selecting the ad, upload your product reference and your guidance.
​
​

And voila!
​

Another example. Turn this template:

Into this by using our own model, wearing our jeans and our shoes (with the collage feature).

Facebook ads templates are just the start. You can expect hand-curated templates that are proven to work across all visual types in the coming weeks and months.

Dezygn Eye Chrome Extension makes it effortless to import images from any website directly into your Dezygn workspace with just one click.

Instead of downloading and re-uploading images, Dezygn Eye lets you capture visuals directly from any website and organize them into your Dezygn workspace. This saves time, keeps your workflow seamless, and preserves image quality.

Now since Facebook Ads are now a thing in Dezygn.
​
The next logical steps would be variations.

Let's say you just created a banging creative.
​
But you need multiple variations of it.

This is now possible in Dezygn using the variations feature.

To create variations, simply select the Quick actions button in the chat area and then "Create variations".
​

You will then be prompted to describe your variations.

Suffice to say, this functionality opens a whole world of possibilities to Dezygn.

New ingredients panel

The ingredients sidebar panel provides quick access to your ingredients at a glance.

Concurrent Image Generation

You can now generate multiple images at once, no need to wait for one generate to end to start a new one.

Multi-File Upload and URL import

You can also upload multiple ingredients at once from your desktop and import images from URL.

Progressive Image Loading

Finally the performance for Dezygn has noticeably increased. Images no longer load in full size but as small thumbnails, which leads to faster loading speeds.

The video studio now includes 3 major new features.

Our fast video generation creates a video in under 10 seconds.
​
This is incredibly full to test your video concept before going full production.
​

Veo 3 is now our Premium Model and can be used to generate amazing videos.

A couple examples of videos I created in the Studio with dezygn.
​
New login page background video.

Fun multi-scene clip:

The latest feature in Studio allows you to have continuity in your video clip.
​
Start your second video clip where the previous one left off.

Also noteworthy, the cost of video generation has been divided by 3 in Dezygn in the past month.

With the updates to the visual capabilities of Dezygn, now creating commercial videos with Dezygn feels incredible.
​
And Studio will only keep getting better.

In the background, I have improved and systemised our support and training.

New Academy video training

For each new feature and release, you can expect new videos added to the Academy.

New articles added to the helpdesk

Each new video is transcribed, summarised and added as articles to the helpdesk here.

Better chatbot

I have been training our chatbot with everything about Dezygn so it's much better at answering your queries.
​
You can access the chatbot by clicking on the settings icon in the sidebar of Dezygn then select "Support". Or you can access it here in the bottom right.

I would like to say thank you for your support, especially if you are on the Agency and Freelancer plan.
​
These plans do allow me to spend more time on development and less time on marketing while I make these improvements.
​
Special thank you to our legacy client and any one that has been making use of Dezygn and sharing their feedback since the latest release.

I plan to post big updates like these maybe once a month. In the meantime you can follow the progress in the Setttings page under Help & Support > Changelogs.

If you are an existing Dezygn client, you can support my work by purchasing our Legacy Special offer, if you haven't done so yet.

Get the Dezygn Legacy Special

You can also subscribe for more credits and access to Agency features here with a trial.

Try it free