VRChat Avatar to VRM Conversion

Step Two: Drag your UniVRM file into unity and then in the window that pops up select "Import"

Note: It is important that UniVRM is the first thing into the project. Do not download any dependencies before this is done.

Step Three: At this point I recommend placing all of the existing files in the document in a folder called the "Keep Folder". This will just be a place for us to place files we don't want to delete as this process requires a lot of importing and exporting. 

Step Four: Import your VRChat avatar into the project. This can be done the same way we installed UniVRM by dragging it into the project and selecting "Import".

Step Five: Drag your model into the scene and ensure all blend shapes are set to 0. These are often stored on the main mesh of the model. We will be reapplying customizing blend shapes in the VRM Blend shape creation.

Step Seven: In the window that pops up you can ignore most things and fill in the required info of "Title", "Version", and "Author". You can choose to fill in the other information as well but it is not required. The title and version you can fill in how you wish. For Author please ensure to do your best to credit everyone involved with the process.

Step Eight: Click "Export" and save the file somewhere you can find it. I renamed the file to the Title and Version, this is not required. Perhaps even save a backup copy of this file so you don't have to repeat the previous steps.

Step Ten: Drag your UniVRM file into unity and then in the window that pops up select "Import"

Note: It is important that UniVRM is the first thing into the project. Do not download any dependencies before this is done.

Once this is done you can close the project and install dependencies you desire like Poiyomi (*Link*)

Step Eleven: go to the VRM0 menu at the top and select "Import VRM 0.x...". Open your saved VRM file from earlier.

This will pop up a second file to name the prefab that is created in this process. You can name this file what you want but I usually keep it the same name. Click "save" and by default it should save into your "Assets' Folder.

Step Twelve: Drag two copies of your prefab into the scene. One of these is going to become the main model we are working with. The other will act as a posing dummy so we can double-check that the blend shape values we assign are to our liking.

Note: On our model set to be exported blend shapes will be applied through the VRM blend shape set-up, the same way we will be handling facial expressions. I have found this method works most consistently and gives the end user more flexibility when it comes to having presets to modify their avatar appearance on the fly.

Step Thirteen: Rename the (1) copy of your avatar to "Posing Dummy"

Step Fourteen: When you imported your VRM this created a folder that is the name of your file and "Blendshapes". Open this folder

Inside the inspector for this Blend shape asset we will see a list of all of the other blend shape clips in that same folder, highlighted in Red. This is were we will be editing all of these files from. The currently selected blend shape is below this, highlighted in Purple. Conveniently, all of the model's blend shapes should be listed at the bottom, highlighted in blue. The blend shapes in blue will show to what value the model will be modified as the blend shapes in red are actived.

Step Sixteen: Click the "Create BlendShapeClip" button

Step Seventeen: In the window that pops up name the file "Body Modification" and click "Save".

Note: As a practice, I do not edit the normal blend shape clip as I want fine control over the model in VNyan and having a "0" point can be useful in testing and troubleshooting. You can create multiple blend shape clips for different body modifications using this method. This lets you have for example a "Chubby Body Modification", or a "Feminine Body Modification". You could also split these variations into different VRM files if you want to keep it more simple.

Hint: You can press the "Lock" symbol in the inspector to keep it on this menu. That way as you are working with different parts of the project you can always have this window up regardless of what you have selected.

Step Eighteen: On your Posing Dummy right click the part that contains your normal Blend Shapes and select "Properties". This should pull up a moveable window that you can keep up and edit the shapes at the same time as the Blend Shape Clips.

Note: Blend Shape Clips will be applied to the main avatar once in VNyan but we set them up in Unity and preview what they will look like using the posing avatar. It is important to set up major body modifications first as they may change the appearance of facial animations. The appearance of Blend Shape Clips may change slightly between Unity and VNyan, due to them being slightly different environments, and those changes are model dependent. Trial and error is important here. 

Hint: If your Blend Shape Clips have multiple of these tabs you likely have blend shapes spread across multiple locations on the model.

Step Twenty: Next select the "A" blend shape clip. We will be going through these sounds one by one as these are used to handle speech derived animations. The main ones are listed are "A, I, U, E, and O".

Step Twenty-One: Under the "A" Blend shape clips if you go to your model's blend shapes there is likely already a "aa" or "ah" blend shape. This is used in VRChat too to handle mouth and face movement based on sounds from the microphone. For "A" turn on the "aa" or "ah" blendshape as appropriate. 

Note: You can preview what this looks like on your posing model by increasing the value of the same blend shape. This will help you determine how much you want this slider to trigger but is not always a 1-1 representation in VNyan and will take some trial and error to adjust to your preferences. You can also add other blend shapes here to support head and ear or similar animations depending on how complex you want your speech to be represented on your model.

Step Twenty-Two: Repeat this process with the Blend Shape Clips for "I, U, E, and O" using whatever sound is closest in your model's blend shapes. For "I" it may be "ih" or "ie". For "U" it may be "ou" or "oo". For E it may be "eh" or "e". For O it could be "oh" or "o". Checking these on your posing dummy to ensure the facial structure is right for the sound you want represented is important.

Step Twenty-Three: By default there are 3 blink blend shape clips. I personally only use Blink_L and Blink_R as in VNyan your camera can track each eye independently. I sometimes still set up Blink as a toggleable option. But with only the Left and Right blink configured you can still close both eyes together IRL to achieve a blink, wink, or closed eye expression. For this clip your model usually has a blend shape with similar names called "Wink", "Eyes Closed",  or similar descriptors that you can increment. I find this is another area were Unity can look slightly different to VNyan so trial and error is important.

Step Twenty-Four: For our "LookUp", "LookDown", "LookLeft", and "LookRight" we will follow a similar idea to the Blink blend shape clips. Some models may have one or more blend shapes to help add detail and expression to these movements.

Step Twenty-Five: At this point we can select our main avatar in the Hierarchy and use our VRM0 menu to export the VRM file. Please reference steps Seven and Eight here.

Step Twenty-Eight: In the inspector in the top right you should see the material you have selected. Below this there is a drop down menu for "Shader". Click on the drop down.

Step Twenty-Nine: In the menu that appears search for "MToon" and once you see it under "Shaders", click on it. Make sure to change all materials to MToon or another compatible shader.

Step Thirty: For transparent materials in the new menu "Rendering Type" needs to be set to "Transparent". In my example the eyes have a transparent layer.

If you have custom textures for materials you place them in the "Lit Color, Alpha" and "Shade Color" slots.