Feature write uint8 uint16 uint32 float32 typedarrays #366
Conversation
|
I juste leaved an issue regarding 1-Bit files (UInt1). Would it be possible to add this type to your work? |
| export function isTypedUintArray(input) { | ||
| if (ArrayBuffer.isView(input)) { | ||
| const ctr = input.constructor; | ||
| if (ctr === Uint8Array || ctr === Uint16Array || ctr === Uint32Array || ctr === Uint8ClampedArray) { |
There was a problem hiding this comment.
good catch also including the clamped version of Uint8!
| width, | ||
| }; | ||
|
|
||
| const newGeoTiffAsBinaryData = await writeArrayBuffer((originalValues), metadata); |
There was a problem hiding this comment.
I think you have an extra parens wrapping originalValues
|
|
||
| const samplesPerPixel = ifd[277]; | ||
| let elementSize = 4; |
There was a problem hiding this comment.
could you add a comment describing why you set the default to 4?
There was a problem hiding this comment.
could we consider defaulting elementSize to 8, so we can support all floating point numbers by default?
|
|
||
| forEach(values, (value, i) => { | ||
| if (!TypedArray) { | ||
| data[numBytesInIfd + i] = value; |
There was a problem hiding this comment.
I think this would only work if the Array of values is within the 0-255 range. Is there any way we could update this to support Float64 by default? Could we do view.setFloat64 to support any floating point number?
Open to your thoughts!
| metadata.StripByteCounts = [numBands * height * width]; | ||
|
|
||
| // default for Uint8 | ||
| let elementSize = 1; |
There was a problem hiding this comment.
I'm wondering if we want to change the default to 8 for Float64 instead of Uint8? Ideally, the writer will work for the greatest number of use cases by default. If someone wants Uint8, they could pass in a Uint8Array of values instead of an Array.
|
@daumann and @KevinDonnot , I would love to see 1-bit support soon, but I would recommend doing that in a follow up PR. |
No description provided.