Draw Things: AI Generation 17+

Create & Edit Images with AI

Liu Liu

    • 4.3 • 33 Ratings
    • Free

Description

The fastest local image generation now with on-device LoRA training! Combining SDXL with full-suite of image generation / editing tools right in Draw Things!

Draw Things is a AI-assisted image generation tool to help you create images you have in mind in minutes rather than days. Master the cast spells and your Mac will draw what you want with a few simple words. Draw Things runs locally on your phone to protect your privacy.

What’s New

Version 1.20240416.0

* This fixed a bug where using Tiled Diffusion with CFG > 1 or no-accelerated LoRA can generate undesirable images.

Ratings and Reviews

4.3 out of 5
33 Ratings

33 Ratings

DragonMist777 ,

Great App

Draw Things is a great app to help create artistic resources, resize and upscale and make interesting image ideas. Highly recommended.

One problem can be when on a slow connection and downloading that I had to re-download so a resume feature would be useful since the downloads can take a while. The modal window for downloads can also be slide downwards and accidentally be closed if you tap out of it.

In regards to downloading and in general the split screen view would be useful because it could allow to do something else whilst a download takes place. Also this could possibly be used for drag and drop loading of photos perhaps.

A feature that would be really amazing however is the use of Apple Shortcuts (Intents) where you could get/set prompts, change parameters, randomise the seed value and render images to the app or photos as a image file etc. It would also be great to be able to input a image this way from another app or photos using Shortcuts and then upscaling it with a selected algorithm using Shortcuts, then save to an image file.

Having URL Scheme options to launch with images etc would also be really useful to have available.

Mind's 'I' ,

Excellent app

Difficult to find words which adequately express my appreciation for the dedication and talent that has clearly been applied to the creation of this app. It gives the impression of being a 'labour of love', and I wish that I had the talent to help in its development.

I've found adding additional models from outside sources to be a bit of a challenge, but that is most likely to be due to my limited knowledge. However, this app seems likely to inspire anyone with an enquiring mind, and child-like curiosity to persevere through the learning process which the depth of this beautifully crafted app presents.

I can only hope that Apple appreciates how lucky they are to have been favoured by the development of such a creative instrument, which if it gains the publicity it deserves, could be a significant motivating factor in the purchases of iPhones and the recent high performance iPads.

Sincere grattitude to the developer for making this unencumbered beauty freely available to all those who have a creative urge, but are not as gifted with natural talent as artists themselves. Thank you LouLou.

iandol ,

Fast and Featureful!

I've been using imaginAIry from the command-line so far, but the release of Diffusers from hugging Face made me have a look at stable diffusion GUIs. Of all the options, this is the ONLY one that combines a rich feature set with the speed of CoreML. You need to hop onto discord to get help, but the UI does provide quite a lot of hints. I don't know how the developer supports CoreML so well, most tools either use generic models OR converted ones, this is the only one that seems to handle either! A great tool!

Developer Response ,

Thanks for the kind words! We support CoreML 1.x models at 512x512 resolution and implemented something called swift-fickling to support pickle format checkpoint files. By reverse engineering CoreML's mlmodelc format, we can generate proper mlmodelc file on-fly from the pickle format checkpoint. Other resolutions and 2.x models are supported through Metal, which is fast enough and quite memory efficient. We will implement 2.x model support in CoreML in the future which should have 50% speedup over the current implementation!

App Privacy

The developer, Liu Liu, indicated that the app’s privacy practices may include handling of data as described below. For more information, see the developer’s privacy policy.

Data Not Collected

The developer does not collect any data from this app.

Privacy practices may vary based on, for example, the features you use or your age. Learn More

You Might Also Like

RealityScan - 3D Scanning App
Photo & Video
Deep AI - The AI Art Generator
Photo & Video
DaVinci Resolve for iPad
Photo & Video
Affinity Photo 2 for iPad
Photo & Video
RunwayML
Photo & Video
Photomator
Photo & Video