Follow-along Guide on InvokeAI, it's Canvas and making Multi-Character Images!
Compared to my LoRA Guide, which if you still haven't checked out, here is the link:
https://civitai.com/articles/6990/comprehensive-multi-outfit-lora-guide-by-khanykov01
This guide will mostly show you how I make my images and what tools I use.
Same as last time I've made a video guide for it, but most of it is just watching me make complicated images for 1.5 hours... So if are interested in the entire process feel free to watch it.
As for the written version it's more grounded and will just cover the tools and some steps I would usually take to my complicated images.
!Now a quick Warning!
The methods/tools showcased in this guide might not work in other Ai generators like ComfyUI/A1111/Stable Diffusion/Etc. as I have no idea if they have the same tools.
If you are using a Web Generator like CivitAi/Tensor.Art/Seaart, then this guide won't help you.
And as per usual: THERE IS NO TL;DR, Read/Listen properly.
Don't be scared by the video being almost 2 hours this time though, most of it is just me sitting making a few images with multiple characters.
Yes it takes a while, it is not a 5 minute job, if you thought it's done quickly, you can leave now...
Since Youtube doesn't let me put any Links in the description, here is the link to InvokeAi's github and their official website where you can download the program itself.
InvokeAi Github:
https://github.com/invoke-ai/InvokeAI
(Mainly leaving this here for those interested in reading about it)
InvokeAi Official Website:
(Download from here, it's more straightforward)
By the way I have no clue why Table of Content Put Intermediate and Advanced Level at the bottom even though they still lead to the same part of the guide... Thanks CivitAi...
LoRA Guide Start:
Levels and Categories:
Use Table of Contents on the Right part of your screen if you are looking for a specific part.
Beginner Level:
0) Specs needed to use InvokeAI.
Learning how to install InvokeAi and using it to generate your first Images.
1) Installing InvokeAi
2) InvokeAI UI and Model Importing
3) Generating your First Image
4) Image Gallery
5) Settings/Hotkeys/Downloading
Intermediate Level:
Time to learn the main part of InvokeAi: Canvas!
1) InvokeAI's Canvas
2) Using Canvas and Inpaint
3) Character Cutout Addon
4) First Multi-Character Image
Advanced Level:
1) Background/Character Extensions
2) ControlNet (Depth Type)
3) Interactive Multi-Character Image
Beginner Level
Let's Talk Specs before we get to Installing anything, shall we ?
First and foremost I hope you've already understood that InvokeAI is a PC Program, so you obviously will need a PC or at least a Laptop.
Now since we will be covering mainly SDXL/Pony/Illustrious Models, the specs are not going to be too crazy.
To clarify I am basing this off what I have, I don't exactly know what works best or if it even works.
My current Specs:
GPU: GTX 3060 (12GBs)
CPU: Ryzen 5 5600X
Storage: NVME M.2 SSD
RAM: 16GBs
Assumed Requirements:
AMD/Intel GPUs might not work due to a lack of Cuda Cores, so I can't guarantee that they will work. (InvokeAI does have an option for AMD, but I would assume it doesn't work as well as Nvidia GPUs)
Overall 8GBs of VRAM is going to be your base line for not a snail-pace generation.
SDXL Models are mostly 6-7 GBs, as long as that fits into your GPU's VRAM it should work without any drops in speed.
12GBs+ is going to be the recommended amount of course.
Nvidia GPU Series:
10xx - Should Work
20xx - Good
30xx - Awesome
40xx - Amazing
50xx - Perfect
Storage:
InvokeAI itself is 50GBs and your Models + LoRAs will take up around 10-20GBs of space.
You will be able to download images and move them to another disk, so no need to worry about that space.
SSD is preferred, but HDD will work as well. The launch and download speed are probably the only affected parts.
RAM:
16GBs is more than enough, the more the better of course as you can later turn on Low-VRAM mode, where it will use your RAM as a backup of sorts.
In case you are interested in trying to use it for something like FLUX then here is InvokeAI's guide on how to turn it on: https://invoke-ai.github.io/InvokeAI/features/low-vram/
I'd rather not explain how this works myself as even though I've enabled it, it's a bit too technical for me to talk about...
Got the Specs ? Let's go install InvokeAI now!
1) Installing InvokeAi
Pretty self-explanatory to be honest.
Either Google "InvokeAi" and head to the official website or just click on the link to it:
Now simply download it for Windows. (I have no idea how it would work with Mac or Linux btw)
After it has finished downloading the installer, simply launch it.
You'll get a window, which should look something like this.
(Mine looks slightly different as I have it installed already)
Now simply choose what disk/folder you want to install InvokeAI to.
Click NEXT when you've chosen your folder.
It will then tell you the latest Stable Version, as of 05/02/2026 it's v6.11.0. Then click NEXT again.
Now it will ask, what GPU do you have. In my case it's the (30xx and Above) option.
In your case it might be different of course.
As you can see there is AMD option and No Dedicated GPU.
No clue how both of them work, so you are on your own there. :p
And that's pretty much it, InvokeAI will give you a review of what you've chosen and you can click the Install Button. Now since it's your first installation and not an update it will take some time to install. But no worries progress bars and numbers are included, so you'll know that it's working!
After it has finished downloading and installing your application should look like this.
But before we launch please take a look at the Gear in the Top Left Corner.
We mainly want to look at the first option. If it's check-marked InvokeAI's UI will work in your Browser as just another Tab. (This is what I use)
If it's not check-marked InvokeAI's UI will work as a standalone UI on your taskbar.
There is no difference, but as I said I prefer the Browser version, so I will be showing you that.
Go Ahead and Press Launch.
2) InvokeAi UI and Model Importing
After pressing launch InvokeAI's console will take a few minutes to set everything up.
After it is done you should see something like this.
It also shows you the IP address you will use to connect to it in the Browser.
You can either use 127.0.0.1:9090 or 0.0.0.0:9090
Though to be fair the one with 0s never worked for me, so I prefer the regular 127 self-IP.
Welcome to InvokeAI UI!
Let's get started by going over what is where.
On your Left side of the screen you have everything related to generating your Images.
1) The Invoke button is to start Generating Images + Amount of Images you want in a batch
2) Aa Identifies that you are going to generate images into Gallery straight away.
As in the generated image do not go into canvas by themselves. Don't bother too much with this terminology for now.
3) Prompt
4) Negative Prompt
5) Image Size + Seed
6) Model (Checkpoints) + Your LoRAs
7) Scheduler
8) Model Manager
As we currently do not have any Models/LoRAs installed let's go to Model Manager and install some models. Click on that 3D Cube in your bottom Left Corner.
Your installed Models will be on the left side, while right side is for getting new models.
1) Your Models
2) Your Imported Models
3) Preloaded Models + Addons that you can install
You can go through 3rd section later if you want, but I'm not going to explain any of it myself as I don't use it.
So go to 2nd Section which is Scan Folder.
Now go ahead and create some folder on your PC, which you will use as either your
Main Model Folder or just an Importer Folder.
Put your downloaded LoRAs/Checkpoint Models into it.
If the In-place Install is check-marked then the models will install in that folder, thus making you a Main Model Folder.
If the In-place Install is not check-marked then the models will be moved from that folder into InvokeAI's Folder, thus making you an Importer Folder. (This is the method I use)
After you choose which Folder you want, click Install on the model or Install All if you have several.
If it installed properly it will look like this at the bottom.
On your Left you can now search for it to see that it was indeed installed.
3) Generating your First Image
Let's head back to Aa to generate our first image.
In the LoRAs/Models section choose your model and checkpoint.
(I will be using my Songque10in1 model, you can download it on my CivitAi profile)
For those curious I use Obsession v3.0 by https://civitai.com/user/rqdwdw
Link to the Model: https://civitai.com/models/820208?modelVersionId=1033430
Now write the Prompt for your model or if you don't want to use a model just write something... I'm fairly certain most of you know how this part works...
Don't forget to add Negative Prompt if you have it.
And click Invoke!
For those Curious it takes roughly 17-20 Seconds for me to generate an image with my 3060 using Euler A at 25 Steps and CFG Scale at 6.
Your GPU + Parameters will of course give you a different result.
4) Image Gallery
On your the right side of your screen you will find your Image Gallery.
Any new image will go into Uncategorised Board by default.
Boards are what we call Albums in InvokeAi.
Now I personally do not like using Uncategorised Board and I like to sort my images, so let's
create a new Board with Plus Sign in Top right corner and rename it to something,
I will rename mine to Songque of course.
When clicking on the board you'll notice an Auto sign appear, that means any new image will go into that Board. Let's now drag Songque's image into Songque's Board.
Simply click on the image you want to move and drag it to any Board you want.
You can also Star/Favourite your images so they appear at the top of the board.
5) Settings/Hotkeys/Downloading
Now let's say you generated a few images and you want to move them all at once
Simply hold CTRL and then click on any images you want to select them. Then move them around if you wish. Boards act similar to folders, so any shortcuts you are used to in moving files will probably work here as well. Such as CTRL on one image followed by CTRL + SHIFT on another image to select all of the images in between 2 images.
Seems like new Assistants have arrived. And I've generated 2 new images with them.
One with Thelema and one with Lantern. (Both are from Honkai Impact 3rd of course)
But I want to generate an image with Songque again, but my Prompt/LoRA are still for Thelema.
This is where Hotkeys come in.
Click on an image to select it then click any of the following keys:
"P" -> Prompt (Will bring back image's prompt and negative prompt)
"D" -> Dimensions (Will bring back image's Size and Aspect ratio)
"R" -> Resources (Will bring back image's Prompt + LoRAs + Scheduler + Checkpoint)
"A" -> All (Will bring back Image's everything. Including the Seed and any other weird addition)
P.S: Sometimes InvokeAi may tell you that it brought Image's parameters back, but it didn't, just click the Hotkey again in that case... It's a bit annoying I know.
Let's talk Downloading now.
You can either download the entire Board by right clicking it and pressing download button.
After clicking the download button a download prompt will pop up at the bottom of the page.
It will give you a Zip of your Images. I personally have WinRar, so it gives me a Zip in that format, yours might be different.
Archiving Board just hides it by the way, you can view it later by going to settings in top right corner and check-marking show archived boards.
And well Deleting the Board will delete everything in it as well as it's assets.
Assets are just imported images you put into the InvokeAi. Try take any image you have on your device and dragging into InvokeAi.
If you want to download individual images, simply right click on them and you'll see the download button. No download prompt this time, it will download it straight away.
Okay and last but not least let's go to Settings tab.
Head to bottom left corner of InvokeAI's UI and click on the gear.
If you wish to see other Hotkeys (We'll talk about some of Canvas related ones in a bit) you can go ahead and view them.
Overwise just head to settings.
Overall I don't change any settings except these 3:
1) If you don't want to see "Are you sure you want to delete this image?" every time, disable it.
2) I don't like seeing progress images, so I disable the entire noise process, so it's just a progress bar under Invoke Button.
3) This one can be left turned on if you are still unfamiliar with some tools, it will give you pop-ups for them. I obviously know all of the things by now, so I disable it.
Now let's scroll down to the Developer Settings.
There are 2 important Buttons you need to know about.
Clear Intermediates
and
Reset Web Ui
The first button will clear up some space, which is used by small bits generated when we use Canvas later. Try to clear it up when it's close to 10k-20k. Overall it's just to clear some space, don't worry it doesn't delete any images.
As for the second button, if your InvokeAi starts to lag at any given point and it wasn't doing that before press the red button. It will clear all of your previous settings so if you were working on something make sure it was saved and then you can bring it all back with "R" or "A" hotkeys.
Those 3 Initial Settings I've shown you will also reset by the way.
That's it for Beginner's Level, so by now you should know how to generate quick simple images similar to what you were doing in a Web Generator, if you've ever used them.
Now let's move onto the fun stuff!
Intermediate Level
1) InvokeAI's Canvas
Let's head to the Canvas by clicking on the button under Aa section.
It should look like this.
Before we even generate an Image on a Canvas, let me tell you how to move here.
Hold Middle Mouse to drag yourself around the Canvas.
Or Press this button ("H" Hotkey) to enable movement:
Go ahead and Generate an image now with Invoke Button.
Straight away you should notice 2 things.
1) The image is not being sent to Gallery.
2) There is some image settings bar at the bottom.
If you continue generating images they will be added to this bar.
Let's go over what each button does.
1) Arrows to move between images. Arrow keys work for it btw!
2) Checkmark will accept an image from the batch and put it on the canvas.
3) Hides the Progress of the images being generated.
4) Saves the chosen image to Gallery (To the Board, which has Auto on it).
5) Allows you to add the image to different layers, but I don't really like using it, so use it or ignore it, your choice.
6) X will discard an image from the bar aka. Delete it.
7,8,9) All correspond to how the images will move during generation.
Moon Icon will not move from the chosen image even if more images are being generated.
Arrow > Icon will switch the viewed image to the one being generated currently.
Arrow + Line Icon will switch the viewed image to the last finished one.
10) Will Delete all of the generated images in the bar.
Now let's accept an image with 2nd Button and make it appear on the canvas.
As you can see Accepting an image does not save it to Gallery as we will probably want to fix it first and then save it, but let's see how to save it first.
There are 2 Options:
1) Save the Entire Canvas
2) Save the Bbox Only. (Right click on your image on Canvas)
Bbox is that frame around our image, we'll get to it in a bit
And it has appeared in our Songque Board.
At the top of the page you can switch between Canvas and Viewer to see your images properly.
Alright now let's actually use the Canvas.
2) Using Canvas and Inpaint
I will generate an image of Thelema with Full Body in the prompt.
As you can see there are imperfections everywhere.
Let's go fix them!
Before we start fixing it, let's see what regular "Paint"-like tools we have here.
On the left side of your Image where you've already seen the Movement Key you'll also find a few other options.
Each Option when hovered over will show you it's Hotkey! It's very useful, so make sure to remember them.
1) Your Brush, you can paint with it. Useful when you are changing things or covering them.
You can adjust the size of your Brush at the top of your screen or
by holding CTRL + Mouse Scroll (I prefer this method)
2) Eraser will eraser anything on the current Chosen Layer.
3) The Rectangle. (I have never used it to be fair...)
4) Allows you to move your Image around the Canvas, we are not bound by the Bbox after all.
Bbox is just our generating area.
!IMPORTANT!
Hold CTRL when you are dragging Bbox or Images for Precise control. Overwise it will be more of a snap-fit, which might not work in a lot of cases, which we will cover later.
5) Well you know this one already, it allows you to move around the Canvas.
I prefer to use the Middle Mouse though.
6) This option allows you to resize your Bbox, we'll need it in the next section.
7) Color Picker, will pick colors for you based on where your mouse is.
One of the Greatest things Humanity has made. Change My Mind.
Now let's learn Inpaint.
Then you generate any image in Canvas in the bottom right corner you will see 2 Layers.
Your Image being a Raster Layer and Inpaint Mask Layer.
But let's pretend the Inpaint Mask is not there, so let's add it.
Click on the Plus Sign above that Denoising Strength section and select Inpaint Mask.
Click on Inpaint Mask and switch to your Brush, you'll notice it is now transparent.
Color in any part you would like to change/improve on your image.
And then Change your Bbox size to be closer to the parts, which you Inpainted.
Here are a few Important things to know before you generate with Inpaint:
1) Inpaint Mask should never touch or go over Bbox frame. It is allowed to go over it if it's at the edge of an image though.
2) If possible make sure Face is always in your Bbox when using Inpaint! If you are generating something far away from Face like Feet or Footwear, you'll need to adjust the Denoising Strength, this part is covered a bit later, so don't worry.
3) The Smaller the Bbox the more details Ai will be able to generate and fix.
With that your image should look like this now.
Before we press Invoke Button, let me tell you about Denoising Strength.
It's quite simple: Higher the Value, Bigger the Changes!
But don't go ham with it, Bigger also means Ai will ignore most of your image.
Use it like this:
0.3-0.4 for Feet/Footwear or things far away from Face, so Face is not in the Bbox
0.5-0.6 for General Fix, from legs to hands to face and of course outfits.
0.7-0.9 for Big Changes, not going to showcase it here, it's shown in Advanced Level.
These are not set in stone so in some cases you can go higher or lower.
I will now set it to 0.6 and click on Invoke Button to generate that Bbox Area with Inpaint.
!Important!
Inpaint is not an immediate solution. Generate several images with it, Choose the Best one. Still got issues ? Remove Inpaint Mask, add a new one, brush over the areas, which are still bad, reduce Bbox Frame, generate again, repeat by slowly inpainting each part of the image until it's perfect.
In my case here I went through 3 stage of Inpainting.
1) First inpaint mask, which I've shown fixed the hands, but the face was still subpar.
2) I removed the previous Inpaint Mask, add a new one. (It's faster to delete it rather than erase it and paint it again) then made the Bbox Smaller. Invoke Button.
3) Eyes were still subpar after 2nd Inpaint Mask, so here comes the third one and even smaller Bbox.
After I got the image I wanted, I clicked blue Save Canvas to Gallery at the top.
Since I didn't have anything go out of the original 1024x1024 Bbox I can save the entire Canvas.
But let's say I imported an image onto canvas like this:
Now I have 2 images on the canvas aka. 2 Layers and the total size of the Canvas is no longer 1024x1024.
So if I were to save the Canvas it would look like this in the Gallery:
So use that only then it's a single image, otherwise use Save just the Bbox.
To reset the Canvas click this button at the top near the Save Button.
It won't reset the Bbox though, so make sure to press "D" on any image with the resolution you want to use.
3) Character Cutout Addon
Just a quick explanation here of what any background remover or Photoshop can do.
Since it's an addon the moment you try to use it, InvokeAi will start downloading it, so you'll have to wait for a couple of minutes until it's done downloading it. DO NOT INTERRUPT IT.
Let's go back to our Thelema image and remove her background.
Click on Select Object in Bottom Left Corner, above Denoising Strength.
You'll get a console of sorts, disable Auto-Process as it's annoying.
Then drag and choose the character you want to cutout.
And click process.
If you don't want to drag it or need to choose something very specific you can click on parts of your image. Green Dots will appear, make most of the areas, but not too many and run Process.
Let it now download the Addon.
After it's done the chosen area will become blue. There are going to be some imperfections, which you try to fix by clicking on them and adding Green dots and processing again.
Or simply ignore that and fix them yourself with a brush and a color picker.
Nothing a quick brush won't fix. If some areas weren't cutout, use an eraser and manually get rid of them.
Okay now you have a cutout of a character without background. You can drag it around and adjust it's size by right clicking on it and then choosing Transform button. Using CTRL let's you move it and resize easier on the go.
Let's move to the last section of Intermediate Level.
4) First Multi-Character Image
Not to clarify this is a Non-Interactive Image.
By that I mean that characters do not interact with each other. They exist parallel of each other
and the only crossover between them is simple layering.
Now it's hard for me to properly explain the entire process and I'd rather not to be fair...
I do the entire Process in the video, so I suggest you go watch it, but I'll give a quick explanation of each step. So let's begin.
Our Goal is to make 3 Characters sit on the same Bench.
(Songque, Thelema and Lantern in this case)
1) Try to make a sketch either in your head or in literal paint of what you want it to look like.
Even a simple Plan is better than No Plan
2) Generate an image with the character you want to be in the middle or in a position, which will give you the best background. We need a base Image. In my case it will be Thelema.
3) Move Bbox away from Base Image, make sure it's back to original size (1024x1024) in my case. Now using the same Background/Lightning Prompt generate images with the 2nd and 3rd Characters in positions/poses you would like and then cut them out using the previously shown Addon. Here are my images with Songque and Lantern after I removed their backgrounds.
(Any weird cutout or brush strokes are because I needed to put them behind Thelema's hair, see 4th point.)
4) Drag and Resize if needed both of these images onto the Base Image and remove/add any parts, which should or shouldn't be there. In my case I had to remove a bit of Songque's and Lantern's legs/hair to make sure they look like they sit behind Thelema.
If needed drag and drop Raster Layers to make sure characters are behind who they need to be
Each Raster Layer also has buttons to push it up or down in the list:
As you can see both Songque and Lantern are behind Thelema's hair.
5) Right Click on Raster Layers of Non-Base Image characters and choose Copy Raster Layer To followed by Copy to Inpaint Mask
This will add an Inpaint Mask, which covers the entire Character
Make sure to remove any inpaint mask parts, which touch other characters!!
Set the Denoising Strength to around 0.5-0.6, Move Bbox to fit the character and run it.
6) Remove the Inpaint Mask and add a new one to fix any small details aka. Face/hands. Use Brash to fix Merged hair with another character and so on and then add weird details to inpaint as well.
7) Then fix other characters. If you've done everything correctly and saved your final Bbox Image, you should have something like this.
Once again if you want to see the Entire Process of how this image was made, go watch the guide video at the start. Go to 40 minutes mark or so.
Advanced Level
1) Background/Character Extensions
This an addon and a bit of a tech.
Let's say you have a character with a specific background and you want to either:
A) Make the image larger, let's say from 1024x1024 to 1024x1444.
B) Center the character by moving the border of the image to side.
Simply move the Bbox to side like this and set Denoising Strength to 0.8 or so.
And once again make sure the character is still in the Bbox when you do this.
Since it's another Addon wait for it to download!
And after it finishes downloading and generating we get an extended image like this.
You can of course go in any direction to use this. Up Down Left Right.
As for Character Extensions, it's basically recolor/brushing over things you want to change to then literally "painting" a rough sketch of what you want. Just don't forget to change the prompt.
So here for example, we generated an image with a prompt: "crossed arms".
Let's change it to "hand on hip, raised hand" instead of generating a new image.
That means we will now take the brush and change the colors of everything that would be in the place of the crossed arms. After that we'll paint the arms/hands where they should be.
Something like this. I colored the center part with colors of the shirt and the jacket.
Painted an arm to go to the hip and then painted a raised arm.
It does look extremely goofy, but now let's add Inpaint Mask with 0.75-0.8 Denoising Strength to the image.
And boom we completely change the image, while also fixing it.
So now you know that you can inpaint any goofy paints of yours into proper arms/hands/legs/etc.
2) ControlNet (Depth Type)
Now when it comes to ControlNet there are a lot of different types.
If you are interested in checking them out go to Model Manager -> Starter Models -> SDXL
And check what ControlNet types are there, download any you'd like and see how it works for yourself.
I personally only like to use Depth Type one as in general ControlNet kills a lot of randomness, which I like to see in Ai generated images, so copying poses from actual images sometimes feel unfair to myself and the people who made those images...
For this demonstration we'll take this image made by *INSERT ARTIST*
Let's say you have an image of your own, go ahead and drop it into InvokeAI so it goes into your Assets. Then drag it onto the Canvas as Control Layer
In the new Control Layer Section, choose your model (Depth Type in this case) and process it.
Again it's still an Addon, so let it fully download!!
Once it's done we'll get a "shadow" of our image.
You can now use this to generate an image, which will use this ControNet as the base.
Let's head to the last Section where we will use this.
3) Interactive Multi-Character Image
You've probably already realised what's the difference between the previous Multi-Character Image and this one.
Previous one was just a few parallel characters, they did not touch or interact with each other.
Now we will make them actually interact and that's where issues will arise 24/7.
To clarify this section alone is 40 minutes in the Guide...
And I ain't explaining it in full here, just quick steps again.
Now let's go over the starting point of such images.
We need a base image, which consists of several characters from the start.
Meaning we need to generate an image, which has characters interact from the getgo.
It's possible to create artificial interaction with the previous Intermediate method, but I'm not going to cover it here as it's mostly the Character Extension, which I've shown before combined with Multi-Character image.
To generate any Interactive image take something as simple as "2girls, hugging,"
As for the actual character make sure to add a single character. Here is an example:
"2girls, holding hands, Songque tags + LoRA"
It can be anything simple enough for Ai to understand as long as you use the same character.
Now let's go full complicated with that ControlNet Shadow we made.
It's Lantern giving Songque a piggyback ride.
We'll remove any parts of the shadow, which do not correspond to our characters and then change the prompt to something close to the shadow. "piggyback, leg grab, 2girls, full body, walking, stumbling,"
Then we'll generate
I accidently lost the original generated image (can still be seen in the video), but
anyways it looks similar to this.
Now straight away I started brushing over parts, which either needed to be overall removed or adjusted to look like Songque's Outfit. Good example: Boots -> Heels-like shape.
Then I inpaint by hand every part where Songque should be, move Bbox closer to her body, change the prompt and LoRA to Songque now and generate it with Denoising Strength of 0.75-0.8. Sometimes 0.9 even.
I then got this image with Songque now in it.
There are a lot of issues that I now need to fix and I also want Songque's arms to go around Lantern's arms, so I will paint them in (while removing her other arm with yellow paint to mask it as her hair) and then inpaint as usual.
Also once I again I really suggest you go watch the guide video on this, it would make so much more sense...
Anyways, after inpainting we get to this.
(I'm literally layering previously generated parts, so it's not a perfect representation of what I had when I generated this).
Regardless... Now that we have the structure ready, it's time to inpaint everything bit by bit and make this image look good.
20 F-ing minutes later... Here is the final image.
Is it perfect ? No I wouldn't say so. Is it good enough where I'm happy with it ? Absolutely.
And for most people this level of Multi-character Image is more than enough anyways...
Funnily enough, but that's it actually... Everything else in InvokeAi you can learn by yourself by simply experimenting and trying out new things. I'm just showing you the basics.
The real fun begins when you know what you are doing and start making stuff like this.
But then again don't expect to be able to just shove so many character into a single image straight away. Take it one step at a time. Since Depth Perception with Ai is screwed over anyways images like this won't work properly anyways... I made this for my Profile Banner, just took eh... Like 5-6 days I think...
Oh and yes Canvas can start lagging quite a lot when you have over 60+ Layers, so be careful...
Also if you want to see more applications of Canvas and these methods check out my Comic if you haven't done so yet! Link: https://civitai.com/articles/21830/comic-dreammatch-honkai-impact-3rd-x-inazuma-eleven
I hope you've learnt something from this guide and can now make images of your dreams! (Please tell me it's not for NSFW...)
I'm also praying that this is the last guide you guys will ask for, I'm tired of making them... It's exhausting...