Rounding issues in getPixelCrop

[MaciejM92]

Hi, good job with the lib! I think I have an idea for improvement as it comes to outputting the pixel values in onComplete callback. Right now it is possible to receive boundaries outside of the original image surface:

1. Use image 1000px width and 666px tall
2. Try to crop with 16/9 ratio, full width
3. Place the crop area to the very bottom

The problem here is that when using Math.round you may end up with y=104px and the height=563px. This is one pixel outside the original surface. I'd suggest using Math.floor and Math.ceil instead. My suggestion would be to use floor for the x and y values while using ceil for width and height in getPixelCrop.

Let me know your thoughts.

[sekoyo]

Hi, Thanks for raising this good point. Your solution sounds good, or alternatively clamping the width and height to remain in bounds as we have the image node (clamp(x + width), imageWidth)). But off the top of my head I think your idea would work too, will give it a think

…

On Mon, 26 Mar 2018 at 11:51, MaciejM92 ***@***.***> wrote: Hi, good job with the lib! I think I have an idea for improvement as it comes to outputting the pixel values in onComplete callback. Right now it is possible to receive boundaries outside of the original image surface: 1. Use image 1000px width and 666px tall 2. Try to crop with 16/9 ratio, full width 3. Place the crop area to the very bottom The problem here is that when using Math.round you may end up with y=104px and the height=563px. This is one pixel outside the original surface. I'd suggest using Math.floor and Math.ceil instead. My suggestion would be to use floor for the x and y values while using ceil for width and height in getPixelCrop. Let me know your thoughts. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#167>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAuZ-pydg2nAVU6xmhe-pty2iRClAtvzks5tiMhCgaJpZM4S66HC> .

[sekoyo]

Should be fixed in 3.0.11

[MaciejM92]

🎉 Thank you!