I was in Vancouver, one of the cities with most of Chinese immigrating population of all Canada and as the Chinese New Year was approaching, all the advertisement and publicity for it started appearing in the streets near the neighborhood where I live, so I took the decision of making a project for this event, but not an illustration because maybe I’ll not complete it on time, for a 3D composition would be a slow process because the machine that I have.
So, I decided to make an Instagram filter and the main question for that was: “How do you make an Instagram Filter?”. To start I made a research about Asian dragons and their characteristics with the shapes, with that base I did the draft of how the project would be and all the layers that I was going to use, in theory it would be needed 4 layers, one for the eyebrows and beard, other for the chin, and other two for the side parts of the dragon head.
Started by making some vectors in based on images from the internet I made the main lines and after that transformed them to 3D models, but under the Spark AR documentation there is a limitation in the size of the final project as the amount of faces used on the models. Because of that I must worked a lot on the 3D models again to optimize them reducing the faces as maximum as I can without losing the definition and shape that I wanted.
When I finally got the best size, shape and faces I uploaded them into Spark AR and they worked perfectly, then I started the process of working with the different assets combining the 3D elements, the materials and trackers to start moulding the project. After some long nights of work, I setup the face as I desired and got the look-a-like wanted.
The 3D models were on place and after some test was working fine, then I started with the interaction a tracking part, for that I passed to the phase of adding the patches that connects the parts of the visual elements with the internal characteristics of each element as the color, size, position and others that helps you to interact from your filter with the users.
Developed all the interaction between the elements, the interface and activities that the user could do, and everything worked properly. After a
lot of test I finally had the filter as I wanted, I developed a marquee for decoration and a banner with the words “happy new year” those in images that were integrated with the filter. After adding some final details, I sent it from Spark AR to Facebook directly to get approval and be online.
I was waiting for the approval for this Instagram filter and I realized the process could take more than I thought. After some days I got a response about the banner that I added arguing there was a lot of text on the Instagram Filter, I asked for a verification but even when the response was
faster it was with the same resolution. Worked on some modifications and sent different versions but they keep rejecting all of them.
Sometimes with the development you must take baby steps, break in parts, isolate what is not working (debugging) and try again, under this thinking what I did was to remove completely the text and send it for consideration, and for the first time my Instagram Filter was approved and online, not as I wanted but it was on Instagram!.
The next step was made a lot of different versions with the text being modified, I tried attaching it to the movement of the head, but it didn’t work. After that I convert it on a 3D model again being carefully about the size and amount of faces allowed by Spark AR for the project. I made some
test and sent it for approval, but it got refused arguing that it was not integrated with the movement of the head. At that moment I was very frustrated to get so many nonsense refusal from Facebook for a filter that for me was clearly following their policies.
Trying to fix this I worked creating other patches for the 3D banner only separated from the others because I wanted only a rotation in the vertical axis (Y) to not be so distracting from the other elements, after finishing everything again I uploaded it from Spark AR hoping it satisfied Facebook policies. But It get rejected again with the same answer about it wasn’t integrated with the head movement.
Finally, I decided because the deadline was coming to close to add the rotation in the horizontal axis (X) but not in the depth axis (Z) and with this last and exhausting version sent it to Facebook, crossing my fingers hoping that my Instagram filter was approved before this year was over.
You can’t relate to the satisfaction that I felt when I got the notification that my Instagram filter was online and available for use for anyone after all this long process to finally get their approval. At the end of the exercise it has been seen by +1200 users and it has been used so was a great sign.
I’m very satisfied with the final Instagram filter and is available for your use from the filter’s gallery or on the following link:
