Street fighter metal Vega 4k howTo

This is a short guide on how to create a 60GB 4k super heavy pixel art animated VG tribute music video from ~scrath.

Besides creating the video I also play drums in this track.

It is a bit painfull until it's done, then it turns rapidly into fame and succes   .-)

Create sprites

To begin with we need sprites to animate. To avoid re-inventing the wheel, cool sprite sheets or pixel art repositories can be a nice shortcut tot take.

However there may be sprites missing. To get VG sprites, I used ZNES emulator.

with numeric buttons you can disable several layers and capture missing sprites.

Besides VG sprites, some new ones from the real world may be needed. In order to do so you can convert actual pictures to pixel art stuff.

Animate models

To animate characters I used photshop. Basically you will have movable parts of the model in different layers and by making some of them visible or not, you will get the feel of movement.

Notice that photoshop is not good enought at rotating pixel art sprites. It distorts the sprites. To do so properly rotsprite is the tool of choice. It isn't really user friendly but you get better results.

Creat the scene

Adobe Flash has been the tool of choice, as it does manage quite correctly layers and objects. We need to import our own transparent sprite sheets and create as many movie clips as animations we want to use.

Having a layer for each movie clip is a nice aproach as it gives visual feedback about what is being shown and when.

Because in this particular case we need to sincronize music with images the speed of the frame renderization is key.

In the 2nd section, music is a 4x4 pattern at 178 BPM and we need to convert it to FPS. If there are different song tempos, a single scene for each will be needed being aware that the value FPS is global to the whole project.
178 BPM /60 (seconds a minute) * 4 (pattern bars in movie clip) gives 11,86^ FPS.
After doing that you may discover that frames in Flahs are no consistent, only orientative. That's why all the frame need to be exported as single images to compile with another software.

For some reasons I encountered render artifacts which I had to solve manually on for each .png.

Edit final video

To convert frames to video, in Premiere has to be done importing image sequence (if you have them correctly numerated it is automatically done). Then in the menu interpret footage select appropiate frameate and it will be compiled correctly.

Apparently not a single video editing software supports nearest pixel scaling. All of them mess the sprites with blurring effects.

Folling this instructions you can use Virtual dub to get bigger same quality video. You will need to create different complete videos for each zoom level. You computer may constantly crash when using 400Gb files. An easier approach is doing everything FullHD and when it is done use Virtual dub again on the result to get your final 4k output.

My premiere file got accidentally lost, so I have no capture to show. Mainly is is a n layer project. Were n is the number of different zoom inputs created previously. In the top layer the pices of zoomed video will finally be the final result.

It is a tough process, and you can get mad at so many software limitations or bugs, but the overall result is worth it !

Cap comentari:

Publica un comentari