polish

drawthingsai · Mar 21, 2024 · 58edd4f · 58edd4f
1 parent dc0fbde
commit 58edd4f
Show file tree

Hide file tree

Showing 8 changed files with 32 additions and 17 deletions.
diff --git a/Documentation.docc/1.UI.md b/Documentation.docc/1.UI.md
@@ -3,8 +3,6 @@
 
 ## First section is about UI of Draw Things app
 
-F_R_O_S_T_Y
-
 
 
 ![First section is about UI of Draw Things app](https://cdn.discordapp.com/attachments/1058172691309404170/1159101377520341042/IMG_0254.jpg?ex=660d4847&is=65fad347&hm=97d9205ef9fd6d7c203ab254a4f60c096b98e58ca8b3dc5eb0dd4c72c7fccb0b&)
@@ -76,3 +74,5 @@ If the words get orange color, it’s signaling it’s more then 75 words, it wa
 
 19 - History. It will show you all history of your generated images and all changes you do for them with other tools. When long press/right click on picture here, you can delete that picture from history. There are the buttons, from left whole history with all edits, middle is tree history showing history from last editing and generation and coffee button for show only generated images. 
 
+
+*Written by: F_R_O_S_T_Y*
diff --git a/Documentation.docc/2.Models.md b/Documentation.docc/2.Models.md
@@ -3,7 +3,6 @@
 
 ## Models importing/deleting/mixing
 
-F_R_O_S_T_Y
 
 
 When you click the model menu, you will find list of prepared models. They are functional and already prepared for usage. Always let the app download all the files(don’t close the app to the background on iOS), or it can cause problems. Models can be between 800MB for 8-bit models, 1,8GB for normal 1.5 models to almost 10GB for SDXL model. 
@@ -23,17 +22,17 @@ Add difference: when you want to make quick inpainting model from other models,
 ![model](https://cdn.discordapp.com/attachments/1058172691309404170/1059117752079630346/IMG_8945.jpg?ex=660974c5&is=65f6ffc5&hm=4bff53bd3acbab8da95800b8a6d449b196ac14fefd394737c803c8c910771403&)
 
 For Lora models, process is almost same, but you import it under Lora models list, not under normal model list. If you want import your own model which is not in model list(there are already models for pix2pix, depth2img, inpainting etc., so you don’t need to import them, choose from the in app list), you must click settings button, after model menu (1), select Customize (2) and here it’s important to read descriptions of all settings. 
-There are three ways how to import custom model. First you need to find website providing them like (https://civitai.com/) or (https://huggingface.co/) or many others. 
+There are three ways how to import custom model. First you need to find website providing them like [civitai](https://civitai.com/) or [huggingface](https://huggingface.co/) or many others. 
 
 After you can choose how to import model in Draw Things. 
 
 ![import](https://cdn.discordapp.com/attachments/1058172691309404170/1059118667721347113/IMG_8946.jpg?ex=6609759f&is=65f7009f&hm=9ae2de67300736500bbd555b363de23a02aa2e743bc4a8db63287d00abf050c9&)
 
-First way is from (https://civitai.com/) there is official support for importing models to Draw Things. If you click the green play button you can directly import it inside Draw Things app. It will open Draw things app with dialog to download it. 
+First way is from [civitai](https://civitai.com/) there is official support for importing models to Draw Things. If you click the green play button you can directly import it inside Draw Things app. It will open Draw things app with dialog to download it. 
 
 Second way is download the model, above you can see how it looks (1) and (👍). You need to see the window for download (2). This is important! If you click download, the file will download inside Files app (or the folder in your macOS where you normally download files from internet). Here, after download you need to move the files inside My phone/Draw Things/Downloads as you can see above. For Mac Downloads folder is there:
 
-###Directory of DT for macOS  
+### Directory of DT for macOS  
 
 /Users/yourusername/Library/Containers/com.liuliu.draw-things/Data
 or
@@ -50,21 +49,24 @@ Draw Things will download the model from link and after you can import it.
 
 After successful import, Draw Things will ask you if you want to delete the downloaded model (not the imported), if you click yes, it will clean Downloads folder for you, so you don’t need to go here and manually delete the downloaded model.
 
-###Changing model name and some details after importing
+### Changing model name and some details after importing
 
 You can download some app for editing, for example Documents by Readdle and go to DT folder, models and here you can find custom.json file. If you look inside, you will see descriptions about models you imported. So if you feel like expert, you can edit them. But for your safety, backup the original file first before editing. 
 
 And small tip if you want to save space. As you can see above in picture from Files app, there is lot of untitled SQLite files. They are created every time you click Start a new project. It’s history files and if you don’t plan to use them, you can delete them all to save space. 
 
 
-###Deleting models
+### Deleting models
 
 If you want to delete models, soon there will be way to delete them from Draw Things app. Now you must visit Files app as you can see in picture above, My phone/Draw Things/Models and here you can delete them. You can sort the files inside Models folder by time when you click the three dots in right top corner. Every model have 2 or more files, one will be under 2 GB and other somewhere around 500 MB. You can delete the pairs and when you will start Draw Things the deleted model won’t show in model list. 
 
-###Model mixing tutorial
+### Model mixing tutorial
 
 If you want to mix models, you must go to settings, open menu for models and click the first button “Customize”. At the bottom you can see two buttons, if you select right one, you will enter the model mixing screen. 
 
 Weighted sum: you select two or more models and combine them together, like Model 1 Generic 1.5 model, model 2 Anything V3. For generic you will choose 0,3, Anything will be set automatically to 0,7. Is mean final model will be 30% from generic and 70% from Anything. 
 
 Add difference: when you want to make quick inpainting model from other models, Model 1 must be inpainting model 1.5 (because you will add difference in this model). Model 2 will be the model you want to add inside inpainting, so for example Anything V3. For Model 3 is important to find base model. It must be the model Model 2 is based (civitai showing this information for most models here in description). So anything is based on model generic 1.5, so Model 3 must be generic 1.5. Now you can experiment for best results, try to set 0,8 for Model 2 and try it. If you don’t like the results, try lower or rise the number and try again. 
+
+
+*Written by: F_R_O_S_T_Y*
diff --git a/Documentation.docc/3.LoRa.md b/Documentation.docc/3.LoRa.md
@@ -3,7 +3,6 @@
 
 ## Everything about Lora and training your own Lora model
 
-F_R_O_S_T_Y
 
 
 Limitation now is minimum of iPad with 8GB of RAM for 1.5 models Lora, use 8 bit models  and network dim not higher then default. iPhone 15 PRO models can crash when training, but you can try it with 8 bit model and network dim set to 8. For SDXL Lora  you will need powerful hardware with lot of RAM.
@@ -18,3 +17,5 @@ Last button is for Lora training. Here you can train your own Lora. Don’t forg
 You can also check YouTube tutorials about Lora training, you will find lot of them and get experience about how to train good Lora, most of the parameters are similar here like in automatic kohya-ss
 
 
+*Written by: F_R_O_S_T_Y*
+
diff --git a/Documentation.docc/4.ControlNET.md b/Documentation.docc/4.ControlNET.md
@@ -3,7 +3,7 @@
 
 ## Everything about ControlNET
 
-F_R_O_S_T_Y
+
 
 Another menu in settings is for control. Inside you can select control you want to use. When you select any control from the list, under it will appear input  and here you will find what layers the controlNET using. If you want to find more about different controlNET, use Google.  Or check the old tutorial section, here you can find few links. 
 
@@ -16,6 +16,9 @@ You can combine different controlNET together like Lora by click plus button bes
 
 You can find controlNET under settings as Control. Read the announcement bellow for more details. Below you can find two links (not because I’m lazy, but clever people then me already wrote some tutorials). Civitai link showing how the models working for reference, you don’t need to download the models here (not all of them must be available in DT) and Github link is for more technical description. 
 
-https://civitai.com/models/9868/controlnet-pre-trained-difference-models
+[controlnet-pre-trained-difference-models](https://civitai.com/models/9868/controlnet-pre-trained-difference-models)
+
+[ControlNet](https://github.com/lllyasviel/ControlNet/blob/main/README.md)
+
 
-https://github.com/lllyasviel/ControlNet/blob/main/README.md
+*Written by: F_R_O_S_T_Y*
diff --git a/Documentation.docc/5.TextInversion.md b/Documentation.docc/5.TextInversion.md
@@ -3,7 +3,6 @@
 
 ## Textual inversion quick tutorial
 
-F_R_O_S_T_Y
 
 
 First you need to open settings and click Textual inversion manage button  at the bottom of settings. Read the description here. 
@@ -18,7 +17,7 @@ You can also emphasize TI. Use it in this format: (<something>:1.2)
 You can use multiple TI prompts together. 
 
 
-##[old version] Textual inversion quick tutorial
+## [old version] Textual inversion quick tutorial
 
 First you need to open settings and click new Textual inversion manage button . Read the description here. 
 
@@ -43,3 +42,6 @@ As you can see, you can write it manually <birb_style> or write only <> and new
 Menu is clever, let you insert only v1 TI for v1.X models and v2TI for v2.X models. 
 
 [demo video](https://cdn.discordapp.com/attachments/1058172691309404170/1070041213064990831/RPReplay_Final1675187416.mov?ex=660c4809&is=65f9d309&hm=d9ae37b23766bbb1684016a1230f54e8f583838ff27ab47c5563766eb0a4dca8&)
+
+
+*Written by: F_R_O_S_T_Y*
diff --git a/Documentation.docc/6.Others.md b/Documentation.docc/6.Others.md
@@ -3,12 +3,12 @@
 
 ## Something doesn’t work
 
-F_R_O_S_T_Y
 
 
 First you can try is to go to settings and click Reset to default  here. After you can go to settings where you will find at the bottom processor icon , when you open it you can click reset to default here as well. After you can start new project  and generate with that default settings. 
 
 Second is reinstall the app, but don’t forget to backup your files 
 
 
+*Written by: F_R_O_S_T_Y*
 
diff --git a/Documentation.docc/7.CoreML.md b/Documentation.docc/7.CoreML.md
@@ -3,7 +3,6 @@
 
 ## Using CoreML for faster image generation
 
-F_R_O_S_T_Y
 
 
 Go to settings and there is processor bottom, click and you can read the description here carefully and check the bottom benchmark’s. For phones, it can be slow or crash the app. Most useful for iPads or MacBooks. 
@@ -21,3 +20,6 @@ Liuliu: There are some numerical differences between DT and CoreML, in particula
 If you use very long prompt or negative prompt (more then 77 words), CoreML can be turned off back to CPU, so generating images will get slow like without using CoreML, it’s limitation of CoreML now. 
 
 DT only uses CoreML (if you enable it) in best case scenarios (512x512, no longer than 77 tokens, no ControlNet). Otherwise it uses GPU like PyTorch (A1111) and other peers. 
+
+
+*Written by: F_R_O_S_T_Y*
diff --git a/Documentation.docc/drawthings.md b/Documentation.docc/drawthings.md
@@ -1,6 +1,11 @@
 # Draw Things
 Documentation
 
+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.
+
 @Metadata {
    @TechnologyRoot
 }