There is an ARKit demonstration named Portal created by Japanese developer Kei Wakizuka. By touching screen, the portal will appear in the real space in the screen with correct perspective and space distortion effect. Through the portal, there is a digital world, but is an immersive one because the audience can also enter the world, and the real world will be behind the portal.

Now there is a variety of similar projects, and most of them are named as Dokodemo Door which is a magical tool in the Japanese manga Doraemon. In the manga, by the Dokodemo Door, Doraemon and his friends can go to anywhere.

Most Dokodemo Door projects are based on SLAM technology, for example, ARKit, the reason is if the audience use Image-Recognization-based AR technology, it is not so easy to enter into the digital world because the audience has to hold the device aiming on the image which will be a terrible experience. But I will keep my experience on both technology, there must be more than one possibility  as Dokodemo Door.

My experiment is based on Vuforia and ARKit, here are two results, and you may see the difference of two design logics:

To create the negative space is the basic step of this kind of AR experience. Now, I only find one method using DepthMask shader. The idea is “a mask object using the Depth Mask shader. This object will be drawn just after regular opaque objects, and will prevent subsequent objects from being drawn behind it.”

1.Based on Vuforia

1.Components

1. A target image, which is only necessary toImagine-Recognization-based AR, and it should be a little bit lower than the hole;
2. A closed 3D object with a hole (I created in C4D) which is used as the entrance for audiences, in my example, it is a cylinder. DepthMask.shader should affect this closed space, to make it as an invisibility cloak to hide other component inside;
3. An inner box with out top side, it will be the most important part, and may have better construction method if you wanna realize a virtual world, but which is not necessary in my example;
4. A ball with ShowInside.shader.

2.Shader

1.DepthMask.shader

 Shader "Custom/DepthMask" {
  
     SubShader {
         // Render the mask after regular geometry, but before masked geometry and
         // transparent things.
  
         Tags {"Queue" = "Geometry-10" }
  
         // Don't draw in the RGBA channels; just the depth buffer
  
         ColorMask 0
         ZWrite On
  
         // Do nothing specific in the pass:
  
         Pass {}
     }
 }

2.ShowInside.shader, here is the documentation from Unity about how to cull front or back (normally, it is default to cull the back, so you can’t see anything inside a 3D object).

 Shader "Custom/ShowInside" {
     Properties{
         _Color("Main Color", Color) = (1,1,1,1)
         _MainTex("Base (RGB)", 2D) = "white" {} 
     }
 
         SubShader{
         Tags{ "RenderType" = "Opaque" }  
         LOD 100
 
         Pass{
         Cull Front    
         Lighting Off
         SetTexture[_MainTex]{ combine texture }
         SetTexture[_MainTex]
             {
             ConstantColor[_Color]
             Combine Previous * Constant
             }
         }
     }
 }

2.Based on ARKit

Components

1. 2 closed objects with holes in the same direction and with different sizes: the outside one should be a little bit larger than the inside one, as the outside one’s function is to hide the inside one by DepthMask.shader, and the skybox material should be apply on the inside one to create a scene with depth of field;
2. A doorframe for the hole to prevent the edge from being too shape. It can be a real doorframe with high quality texture or a black hole with dynamic shader;
3. A scene with several objects to create spatial sense of hierarchy.

There are great examples of mixing animation and graphics, by EYEJACK based on AR technology:

EYEJACK has also released an AR book includes these examples:

I feel it is not such perfect in this period, because of limits of devices and algorithm, but a pleasant way to use AR form in the future, with HMD as Google Glass and etc..

So the problem now is how to achieve this effect with present technology in an easy way that designers with basic coding background are able to do.

At first, I had two ideas:

1. Using shader to make the background invisible.
2. Using a video with alpha channel;

---

The first way works well, I learnt a lot from this tutorial, but the shortage of being tough to locate the video in the ideal place makes me think it might not be a good method. Here is the result:

1. Softwares

1. macOS 10.13.2;
2. Unity 2017.3;
3. Vuforia SDK included in Unity’s installation package;
4. After Effect CC2017.

2. Video

1. In AE, use effect/keying/color range to make the background invisible, and export two videos: one with RGB channel and another with Alpha channel.
2. Create a new composition and put two videos together as this, then render it as mp4

3. AR

1. If you don’t know how to set up Vuforia’s ImageTarget environment, please follow this Vuforia official tutorial, it is quite basic and I don’t want to copy it here.
2. Create a plane in Hierarchy, later it will be used as the platform of video playing, and drag it under the ImageTarget
3. Create a new Material in Project, and drag it on the plane, and also Add Component/Video Player on the plane, and import your video in Project and drag it into the Video Clip which is in the Video Player
4. Create a new Shader in Project, and replace all default codes with these, which I learnt from the mentioned tutorial:

         Shader "Custom/Transparent" {
     Properties {
         _Color ("Color", Color) = (1,1,1,1)
         _MainTex ("Albedo (RGB)", 2D) = "white" {}
         _Glossiness ("Smoothness", Range(0,1)) = 0.5
         _Metallic ("Metallic", Range(0,1)) = 0.0
         _Num("Num",float) = 0.5 

    }

    SubShader {
         Tags { "Queue"="Transparent"  "RenderType"="Transparent"}
         LOD 200

        CGPROGRAM
     #pragma surface surf NoLighting alpha:auto
 
     fixed4 LightingNoLighting(SurfaceOutput s, fixed3 lightDir, fixed atten)
         {
             fixed4 c;
             c.rgb = s.Albedo;
             c.a = s.Alpha;
             return c;
         }
     float _Num;

        sampler2D _MainTex;

        struct Input {
             float2 uv_MainTex;
         };

        half _Glossiness;
         half _Metallic;
         fixed4 _Color;
         UNITY_INSTANCING_BUFFER_START(Props)
         UNITY_INSTANCING_BUFFER_END(Props)
     void surf (Input IN, inout SurfaceOutput o) 
         {
             o.Emission = tex2D(_MainTex, IN.uv_MainTex).rgb;            
             

             if (IN.uv_MainTex.x >= 0.5)
             {
                 o.Alpha = 0;
             }
             else
             {
                 o.Alpha = tex2D(_MainTex, float2(IN.uv_MainTex.x + 0.5, IN.uv_MainTex.y)).rgb;                        
             }

        }
         ENDCG
     }
     FallBack "Diffuse"
     }

5. Select the plane, and change the shader in the material part

4.Test & Build

1. If you have finished that Vuforia’s official tutorial, you should know how to play it with your webcam;
2. If the webcam works well, so you can build it into your mobile device with File/Build and Run;
3. It will automatically run the Xcode software, and this process you can learn from this Unity official tutorial.
4. Done

---

The second way is much easier but it has a shortage that it can only work on apple’s devices because it use Apple ProRes 4444 which is only supported on OSX, check Unity’s official documentation about Video transparency support for more information.

1.Video

1. It is the same workflow as the first method, in the After Effect you will have the result that video has an transparent background;
2. Then you have to export the video as QuickTime format with RGB+Alpha channels
3. On the same window, there is a button named Format Options, click it and in the new window you have to choose Apple ProRes 4444 as the Video Codec
4. Render it and you will have an Unity-Ready transparent video

2. AR

1. Import it into Unity and in the Inspector click Keep Alpha and click Apply
2. Drag it into Video Clip of Video Player Component of Plane which has been mention in the first method, and create a new Material for the plane which shader should be selected as Transparent/VertexLit with Z
3. Done

This app aims to let anyone whether know Thangka or not can learn to draw it.