Draw Things: AI Generation Reviews

I want to preface this by saying that this is ADVANCED AI generation software. I've seen a lot of 1 star reviews of DrawThings by people who don't know what they're doing and are expecting to just type a prompt and get exactly what they want with no/minimal additional tweaking/setup. If that's what you want, use something like Midjourney. However, this software is virtually unparalleled in ease-of-setup and quality for using Stable Diffusion, you just have to do a little bit of learning and then it works virtually flawlessly. DrawThings has support for LoRAs, textual inversion, and additive models. There are also plenty of quality-of-life features (like giving the option to delete models from your downloads folder once you upload to Draw Things.) There is also a wealth of settings to tweak as well as the option to clear the temporary cache to free up space. Underneath every setting is a well-writen and easy-to-read explanation detailing exactly what it does and what you might prefer to use. Massive respect and many thanks to Liu Liu for providing this software completely for free without any in-app purchases or subscriptions, you're brilliant. It's clear a lot of effort and care went into the user experience.

Specifically, on a 4K screen, the right sidebar showing the history shows images at low res scaled up so they appear pixelated compared to previewing the image you click on in the main middle column. Also, on such large screens it would be PERFECT if the grid based layout just shows up in the right sidebar where the history is instead of popping up in a tiny modal screen. I have a magic mouse and for some reason it doesn't scroll with inertia most of the time in DrawThings. It could be the mouse, just that's part of the frustration of the current UI. Another is the inability to move through the history using the arrow quickly. The inability to right click on a history image and do more than just delete it. Would love to see updates to the UI to address these concerns and make it much much much easier to browse thousands of generated images quickly and more easily. There should be a way to end the current sequence of generations, not just skip 1 out of 100. If there's already a way, it's not obvious or intuitive. Also, can we please have an option to generate more than 100? Sometimes I like to leave it running all night. Also, can we please view and manage generated photos even while DrawThings is generating new photos? That'd be great. I'll probably continue to edit this review as I think of new things, so check back often haha. Thanks so much for considering how you will update the UI soon!

This is a completely revised review. The original complained that the program didn't work on Intel Macs in spite of DrawThings Store claiming it did (not the dev's fault). But the devs responded better than my crankiness deserved and updated it to run on Intel macs. I appreciate that.

It should be noted that a big part, perhaps the primary reason, for not being set to run on Intel is that the APIs they're using are optimized for Silicon, so even simple prompts can take several minutes, which isn't a huge problem for casual users. But, it is accessing the models so disk access can get choppy if you decide to watch something whie you wait. Of course, that's not news to Mac users.

The only real issue I've come across is a bug that made me force quit. You need to hold option and click to view tooltips and one time it brought up a color picker that I couldn't get rid of resulting in the need to force quit. That might upset a pro, but I'm good for now.

I’ve fooled around with a great deal of AI Art generators, and none of them have come even close to this. The quality is fantastic, and I like how you can actually download other models and Loras from other sites. Upon using DrawThings, I only have 2 small issues. The first issue being that DrawThings is a bit of a battery drainer. I’ll only have used it for a short period of time, and my iPad battery will have gone from 100% to 90% in…maybe 5 minutes, which is a bit fast, considering my iPad is only a few months old. The second issue, at least in my opinion, is a lack of sampler variety. I like seeing results from the same seed and prompt, but the sampler itself could also produce a slightly altered result, which is one thing I like fooling around with. Hopefully, you’ll add more samplers at some point. Other than those 2 issues, I have nothing bad to say. Nice job.

After having tried several different versions of Stable Diffusion on my 16GB M1 iMac, including the hard-to-set-up Automatic 1111, this is the best by far. Due to superior memory management, it easily runs on my iMac; Automatic 1111, on the other hand, made frequent use of memory swapping that hurt its performance. As a result, this runs many times as fast. It also has far more options than other easy-to-set-up alternatives like DiffusionBee or Guernika, which I've also tried. Finally, it allows you to easily try out multiple compact SD models and also runs well on my 8 GB iPad Air -- an amazing achievement.

If there's any room for improvement, the interface could be a bit more polished and a pull-down menu with full help documentation could be added. Still, these are moot points and this application truly deserves five stars.

Solid work. After struggling with the performance and memory thirst of Automatic1111 on Apple silicon and not getting enough out of the command line CoreML implementations, I decided to give this a try with moderate-to-low expectations. They were surpassed, and how: model setup is as easy as a drop-down menu with pretty much every variant out there, performance is great, memory pressure is surprisingly low, and it does appear to be leveraging CoreML.
Iterative work is very nicely supported by the edit history.
The only glitch I would report thus far is that the coloring menu is finnicky, I cannot do much other than select from the basic palette, which then I can't escape without clicking on 'More colors' (which I don't think is a functioning menu).
Haven't done much with LoRA and control nets yet but so far this thing is smooth as silk.

Most grateful for the Dev for making DrawThings. Thank you.!!! Using on a Macbook AirM1 works very well, and pretty fast.

Easiest install, with the models you'd want readily downaloadable. Great GUI to access the tools and settings you need.

CONS: Hardly any. again so appreciate having DrawThings. It does show it's IOS side, with some things that are not available while using on OSX.

My only wish/suggestions:

1. PLEASE implement a bulk file save for the jpeg option, (rather than having to do it one file at a time...) When creating multiple iterfations, it 's very time consuming and labor intensive to save each picture via right click etc. The work around I'm using now is a macro program, but feels like this could easily be added.

2. Have scroll bars that will respond to keyboard. up/down keys. Assuming it was intended as a swipe thing in IPAD, but on OSX or using a keyboard with Ipad it's not available.

If these minor issues were addressed there is no reason to use any other version of Stable Diffusion- unless you need the most advanced features DrawThings will do iit all. Simple install and works great.

If there is a patreon, or donation page, I'd gladly contribute.

Many thanis again to Liu LIu!!!

I’m on an iphone 13 plus and after downloading DrawThings and waiting a bit to download the 2 GiB system, my device’s photo storage suddenly tripled in size and I went from 68 gbs used out pf 128 to 127.5 out of 128. I proceeded to delete DrawThings and all corresponding files, and reviewed my photo album. I have found no new pictures and nothing to signify the tripling of my photo storage size after the use of DrawThings.

If there is some sort of hidden download that comes along and fills the photo album like this, I can not see it and have gone to apple support about this issue bringing DrawThings in light. I imagine this may be some sort of bug but I have been purging my photos as a result and whatever was downloaded from DrawThings before I even got the chance to generate any photos is still taking up 80 gbs of space on my iphone and I can literally not see it in files or the photos I have where it says it is.

I hope you can offer a solution but I am otherwise highlighting this as a danger to devices with apple support.

This is so cool. I've had this on my phone and iPad and my Mac for a while now, but I'm finally going to write a review. Long story short, DrawThings is incredible and it kind of restores my faith in humanity to see a developer work this hard on an app and not put it behind a paywall.

Granted, I would donate if that was an option - I'm not trying to say I wouldn't pay for the work the developer's done, but that it's free is just remarkable, especially when considering all the ChatGPT apps that litter DrawThings Store with tons of in-app purchases required. But not DrawThings, not this developer!

I'd give it more than five stars if I could. Seriously, this is awesome and anyone who wants to try AI-generated imaging should use DrawThings.

Guys, how is DrawThings sitting at 4.4 stars? There must have been some issues eaerly on in its release because in its current state, all I see is the uncontested king of all free stable diffusion apps.

Its completely local (which freaked me out when it asked to download a 500MB bootstrap on first launch, thought I was about to become a unwilling Soviet spy) and there are zero ads. ZERO. Unlimited use. Tons of tools. Far more than a filthy casual AI art enjoyer could ever fully utilize.

If you are in my class of user: just a person wanting to experiment with stable diffusion AI art without boundaries or "tokens", this is it, this is your app. It even has the full suite of mac, iphone, and ipad apps so you dont have to use an upscaled iphone version on your computer like most other SD apps.

Overall, this is the best AI art generation program I have used, and Ive played with about 20 so far. Even if I find something "better" out in the wild, there is a 0% chance it will be completely free like this one. This is as good as I will get without forking over money and I feel blessed that this developer put so much love into it.

DrawThings enables you to use Stable Diffusion directly from your device with image generation being relatively speedy, usually taking around +\- 1 min. depending on your selected settings and options. Can easily download and use custom models, Loras, etc.

I cannot praise the developer enough, who to my knowledge is the only person to give us an app with this capability, level of privacy, and customization…all FOR FREE. They are constantly putting in work and pushing out updates that not only make DrawThings more efficient, but also provide us with more features and capabilities. I would gladly pay for DrawThings because of what it provides, and also because the developer has certainly earned it.

It’s certainly better than jumping from mid app to mid app, paying their subscription costs and using their network in order to generate like 8 images a day, all the while they are intentionally vague in the descriptions of their privacy policy/granted permissions just so they can double dip and sell your data off to some marketing research firm.

Thanks

Very impressed with the developer's work ethic. Please consider the following:
--My hard drive space is filling up because of adding models, mods, etc. Please add model, mod, etc. management for adding and removing local storage and the ability to have models, mods on an external drive. This way local storage will not fill up. Right now there is no way to save a model on an external and have the software use that model.
--There are a lot of cool features, but I don't know how to use them or how the scale of percentage works exactly. Please add a help, how to, demonstration, etc. Keep up the good work. I would be willing to pay for a version that allows me to get consistent characters in different poses for a graphic novel, comic, etc.

This is a truly well developed app, with tons of configurable settings, such as the ability to train your own LoRAs, to combine models, and to add models other Refiner models. Additionally, the different themes that you can use are really cool. This is also your best option right now for running SDXL, which is a major upgrade from previous versoins of Stable Diffusion.
One small ask, and I don't truly know if this is even feasible, but if DrawThings was able to run .mlpackage files, then I'd be able to run SDXL on my m2 Macbook Air's Neural Engine, instead of the weaker GPU.
Overall, really excellent app, and I'm really really impressed that this is free.

This is straight up the best image generation app - 100% local, no ads, very customizable settings/addons, etc... HUGE props to the developer for keeping this free!

I have only one qualm, on my Mac, DrawThings crashes every time i click generate. it makes it thru maybe 8 or 10 of the little red/blue squares, and then just quits completely. Maybe i have it set up wrong, or perhaps my machine is just too old? (mid 2013 Air)

regardless of that, DrawThings still deserves 5 stars (and it might be my fault anyways lol), this is quite a gem :)

I love this program and thank the developer for creating an all in one solution for the IOS and Mac platform. I can switch using different AI models, but AI art generation is very complex in itself. The outcome is often a hit or miss. Along with the convoluted options and setting that that exist in the program which allows you to do limitless things in itself it's not a bad thing. But if the developer can point the users a more detail documentation or website that would allow the user to understand what each settings does, it'll be a perfect program. Again I thank the developer for his enomourous effort in putting this program together for us to use.

LiuLiu has managed to create an app that most would have said was NOT possible just a few months ago. Not only is it smooth, functional, and intuitive - it also works brilliantly!
Creating AI art on a PC is taxing for most machines, but DrawThings does it on your phone. I’m shocked, dumbfounded, and have had my mind blown by this achievement. I truly hope DrawThings is wildly successful. It’s creator deserves all the credit they are given.

i have been using A1111 for several months and wanted to try something new. This is great for getting out images and getting started. I was able to use SDXL to make images after install and download with out a hitch so major points for usability.

However, when I tried to use DreamshaperXL I could not make images and I think this is due to the fact it didnt know which VAE to use and I couldnt find a place to select the VAE. There are lots of custom VAEs and sometimes I like to mix and match.

The LORA interace isnt great. I think this would be better in tabs and not having everything in line. I have 30 LORAs...but the interface is too clunky for quickly choing a few to use.

It also didnt seem to like that I was using models from another directoy even though the option is there. It would slow down when I selected my models from A1111.

Text size is a little small.

If they can streamline LORA creation on apple silicon this could replace A1111 for me. So far its great and am excited to see where it goes.

Generally works very well. Selection of different models is amazing. Suggestions for improvements:

- Show a preview of what is being generated at each step. This would allow you to cancel if the render is not turning out how you want.
- It’s weird how when you delete an image it is still in the history. There are so many mis-renders I would say if you delete something it should just be gone forever.
- I don’t understand all the little text tags under your prompt. Waste of space and hard to use. Suggest removing in place of a simple clear prompt button.
- “presets” for prompts, especially for negative prompts. Often for characters there are certain things that just make sense to use every time.
- allow you to browse your previously created images while a new one is being rendered. I think the scroll bar is odd to move through renders. I suggest just making a scrollable list.

But very impressive and fun overall!

So i’ve used Midjourney, Nijijourney, StableDiffusion, and Discodiffusion. I haven’t really struggled with this issue. Which is a shame, I don’t know how to remedy this.

I’ve had this issue since I started using this and it has never gone away, which is a real shame as I’d love to use my ipad as a primary source for generations. I’ve tried changing every single setting in DrawThings in various combinations, i’ve tried various models, lora, everything I can think of. But i cannot stop my images from being plagued with rainbow artifact outlines.

Is this a known issue? Is there a work around?

I'm sorry but I was searching for something that might resemble a fully developed image. There are far more powerful apps that you can use on your phone - not your desktop - that return a product with REAL AI accuracy and clarity. DrawThings seems far out of date already. I should be able to say something like "In the style of Stephen King" and DrawThings to understand what I am asking for. Also - they do not tell you that you will also be installing about 4 GIGS of extra "stuff" on your computer in order to run "with priavcy." I don't know where that 4 GIGS of stuff is right now, but I am searching for it and rremoving it as quickly as possible. App is useless in 2023.

It's fast, free, updated all the time, and developed by one of the most resposive, nicest devs I've seen on DrawThings Store. The only downfall here is I find the UI to be clunky and difficult to work when compared to A1111's gradio UI (even when it's frankensteined together with 11ty extensions).

I'm missing the functionality of the A1111 extension ecosystem. Some documentation on developing scripts for DT would be very much appreciated. I would use DT more if it had something like the XYZ plot script available. Only one script (as of 27-nov-2023) ships with DT and it's not a port of that. I'd love to try and whip up a port of the xyz plot script but would be much more inclined to actually do it if there were a website somewhere with examples and documentation.

AI image generation and interpretation is complex, and there is a learning curve with the user experience (UX). THe "cheat sheet" text near the controls really helps the user to understand the basic purpose of the controls. There is a lot to learn here. I've used this and automatic1111 for generating images locally, and this one is better in my opinion over automatic1111 especially for begiiners. There are still some improvements that can be made to the UI (like how to delete images from history feels awkward and clunky, but it is serviceable). Overall, great job on this Liu Liu, and thanks putthing this high quality offering out for free!!!

DrawThings is great for offline ai generation, I loved using it but unfortunately I have an older phone so it isn’t very fast, DrawThings has to be kept open the entire time an image is generating but it isn’t really that bad. HOWEVER, because my iPhone is so old it heats up a lot and using DrawThings seems to have severely affected my battery health, my phone can’t even last the whole day now. Please understand that I am using an older phone with a processor that isn’t the best, I know DrawThings would be much better on a newer device. if you are using a newer device then please disregard the above message. DrawThings is great, I love the concept, but if you have an older phone it’s probably better to download AUTOMATIC1111 on a pc if you do have a good GPU, If neither of these are good options for you it’s probably best to use an online image generator like civitai, account are free and you get lots of images per day (around 200), but if you don’t want to deal with an image limit and you have a good phone or PC then DrawThings is the way to go.

Devs have been consistently updating since release to keep app ahead of the curve. It's got a great UI, but it is a little intimidating at first. Best practice is to not change anything you're not familiar with. Most options have a brief (and collapsible) explanation. It runs great. I have nothing negative to say about DrawThings.

All that being said, if the devs read this, I do have a few reccomendations that I think would be well received. I would love to see an option within model and lora management to add notes about them; with models getting more and more diverse with varying requirements, being able to add "directions" somewhere would be really convienent. In addition to that, it would be nice to be able to rearrange thier list order and to rename them after importing. Lastly, please please please give an option for removing the preloaded models, loras, etc.. It gets pretty crowded once a lot of models get imported and all the place holders for models I'll never use is a little annoying.

Lastly, and people are going to hate me for this, but I think y'all should strongly consider running NON-INTRUSIVE ads across a banner at the bottom or top with a couple dollar paid version to remove it. I don't know how DrawThings is monetized, but you need to be getting paid.

Good luck, thanks for a great app, and thanks for the constant updates!