Stable Diffusion on iPad - Draw Things Free Stable Diffusion App for iPad & iPhone

Key Takeaways

hey what's up coders welcome to one little coder in this table diffusion tutorial I'm going to show you how you can use table division just within your Apple iPad yes if you have got your Apple iPad you can use table diffusion completely locally within your iPad nothing is connected to Cloud you don't have to upload any data to any cloud or any APA everything happens on your local iPad and I'm going to show you which app you can use and how can you use to create your stable diffusion images as you can see right now this is the interface this is from an application called draw things and when you create a create an image using stable division this is how it is going to look like and I'm going to show you how you can do it for your own iPad the first thing is you need to go to the App Store and search for an application called draw things search for an application called draw things on app store and once you have it I have already installed the application so it shows open but if you are just opening it for the first time it would show install here so install it it would take a lot of time for you to install one the second thing is the when you open the application for the very first time it would also take a lot of time because it's downloading all the required models so I'm going to click open it this is assuming that I've already done installation this is the interface that you would see for stable diffusion on iPad if you are familiar with stable diffusion you know the basics here so I'm going to quickly go over all the basic UI components here the first manual item that you have got is the model that you can select and once you click that you can see what are the models that are available for you to use and what is the model that has been already downloaded in my case stable diffusion 1.4 is downloaded on the other models you can see that it shows the download option there are a lot of custom models also available so you can use it if you want so I've stuck with this table division 1.4 the next one is the seed value here you can edit whatever seed value you want the next one is the image size so 512 by 512 is the default value but if you want to use any other image size you can change it do you want an upscaler installed once again upscaler will have to download respective model so you have got a multiple upscaler options um especially the most popular one easier ESR Gan is available and also you have got Universal upscaler so if you have got upscaler enabled one your generation will take time because it has to upscale two it is also going to occupy more memory on your iPad because it's going to download a new model and the next thing is it's going to show you the final output size and the next thing is you have to select the number of steps number of steps will help you get the image um like probably better than what the Lesser number of steps would be and uh that's that's number of steps is closely correlated with the image quality so that is number of steps more more importantly it is going to try to align your image more with the prompt so that's number of steps the next one is guidance skill guidance scale could be anything from zero and it can go up so I have always found guidance kill 7 to 10 working fine so this is the same thing that a lot of UI is called CFG the next thing is strength this is applicable only when you do image to image I forgot to tell you that you can also do image to image with this not just text to image the next thing is the next very good thing is you can select Samplers at this point you've got four Samplers that you can select with all popular ones DPM plus plus Euler ancestral plms ddim so I have found best results using Euler or DPM plus plus 2m carers so you can select whatever you like the next thing is this interface is where you give the positive prompt and this interface is where you give the negative prompt while I'm going to show you the rest of the items I'm going to just give a very simple positive of prompt and let's see so I'm going to say a a bear working with the computer on a sunny evening and that's it I think this is this is fine and negative front I'm not going to give any negative prompt I'm going to click generate the moment you click generate you can see that it is going to give you certain options it says that this is you are trying to do text to image generation and number of steps that you have got guidance skill that you have got what is image strength and also the sampler and as you can see the blue color ones or the progresses that are made the orange color ones are where you know it is trying to make progress assume this is like a waffle chart and this is your progress bar that will give you every time you generate an image you would get to see it in the right Gallery which you can also clear up after your usage so this is the entire UI if you want to save an image you can save an image here if you want to share your image you can share your image if you want to increase the number of batches if you want to increase the number of image you can do that clicking here and you have got certain presets that you can click here and these are the options that are related to history image in painting and all those things are applicable here so these these these circular icons that you see here that that can help you with those things but overall this is a really good interface and this is completely local my iPad is not even a M1 iPad I think it is A5 bionic I don't remember the process but it still works fine the only catch that you need to remember is iPad heats up a little bit as you use you know that this is a computation intensive work what we are doing is so iPad heats up a bit and right now I'm not going to edit the video so you know exactly how much time it takes in the Real Time video like for example you can go back to in the video and then see when I click generate and then you can then see how many seconds or how many minutes it takes for every image generation I think in my estimate takes about two to three minutes that also depends upon the kind of model you've got okay cool we have got a bear working with a computer on a sunny evening I don't know Sunny evening here but at least we have got a bear working with a computer and if you want to try different images then you can again once again change the resolution download a new model add a negative prompt I mean basically everything that you do with a UI like invoke UI automatic 11 11 you can basically do it and everything happens here on your local computer I mean iPad I don't know whether iPad is a computer or not so everything happens on your local iPad that means you have entire privacy about what you want to do and you can upload your own personal images to in painting do output this this is to move for out painting the next one is this one you can erase a particular part the next one is for you to do again to help you with inventing if you want to delete delete this if you want to take a picture I mean you can explore all these things this is quite intuitive it is a free application so kudos to the developer who has made it and I've made it available for us to use all I am really excited about the possibility that you can use state of the art image generation without internet I mean you don't need internet after you have downloaded the model you can literally switch off your internet and still you'll be able to use it because everything happens locally so this is mind blowing for me I hope you would have also enjoyed it if you have enjoyed it please make sure that you go to the App Store give a kind nice review to the developer it would mean a lot to them give a nice rating give a nice feedback if you have got any feedback but otherwise I hope you enjoy this new iPad stability stable division UI on iPad that's completely free running completely on your local iPad if you have any questions let me know in the comment section see you in the next video