[bug fix] odd dimensions cause encoder issues #95
Conversation
- video dimensions cannot be odd in MP4 and many other formats. Force an even scaling factor and dimension, which will always result in an even resolution in the final video.
| @@ -54,7 +55,7 @@ export class StageMap extends Component { | |||
|
|
|||
| _resizeVideo() { | |||
| const {width, dimension} = this.props; | |||
| this._setDevicePixelRatio(dimension.width / width); | |||
| this._setDevicePixelRatio(nearestEvenFloat(dimension.width / width)); | |||
There was a problem hiding this comment.
what's the reason behind using Float here and Int above?
There was a problem hiding this comment.
The ratio is a scaling factor and uses float (e.g. 1.345), whereas the dimensions represent pixels and those should be int (e.g. 1080 or 720).
| @@ -0,0 +1,7 @@ | |||
| export const nearestEvenInt = num => 2 * Math.round(num / 2); | |||
There was a problem hiding this comment.
what's the reason for diving num / 2 and then round and multiplying 2?
There was a problem hiding this comment.
The function needs to return an even number, in this example we can see how this works:
Math.round(13) = 13
2 * Math.round(13 / 2) = 14
Pull Request Test Coverage Report for Build 919445446
💛 - Coveralls |
| export const nearestEvenFloat = num => { | ||
| num = num * 1000; | ||
| num = 2 * Math.round(num / 2); | ||
| return num / 1000; | ||
| }; |
There was a problem hiding this comment.
What's your thought on making this function takes in decimal place rather than making it fixed at 1000?
ex)
export const nearestEvenFloat = (num, decimalPlaces)
There was a problem hiding this comment.
we can also do something like
factorTen = Math.pow(10, decimalPlaces)
There was a problem hiding this comment.
Oh I like that. That makes me think we can just make a nearestEven function and use 0 for the int case.
Video dimensions cannot be odd in many video formats. Solution: Round scaling factors to the closest even factor to always get an even resolution in the final video. Any number multiplied by 2 is even.
The ratio is a scaling factor and uses float (e.g. 1.345), whereas the dimensions represent pixels and those should be int (e.g. 1080 or 720).