I am trying Photosounder 1.90 demo and here are some comments :
The synthesis engine is very good. One of the best , if not the best, I have seen for WAV to PIXEL.
I think you did the good choice of mapping from Db to PIXEL value . I suspect a wide DB range for that quality like that … around -60db to 0.0 db i will guess , maybe more…
I like also the operation panel like horizontal average, vertical average, the mapping by square,
1/x , and square root or log… I would have add convolution kernel 3x3 and 5x5
to be able to do blurring, contrast, edge detection or punch reinforcement … the idea
of going back and forth with photoshop or Fiji imageJ …. hum….
There's only one thing I dont understand: Why there is no Selection Tool. For
exemple if I want only save from 2sec to 4sec how can I do it…
Best regards,
qcmv22
Photosounder
-
- KVRian
- 1050 posts since 6 May, 2008 from Poland
Oops, I missed this message entirely. I didn't map decibels to pixel values because unlike every other programmer I'm not a complete idiotqcmv22 wrote:I am trying Photosounder 1.90 demo and here are some comments :
The synthesis engine is very good. One of the best , if not the best, I have seen for WAV to PIXEL.
I think you did the good choice of mapping from Db to PIXEL value . I suspect a wide DB range for that quality like that … around -60db to 0.0 db i will guess , maybe more…
I like also the operation panel like horizontal average, vertical average, the mapping by square,
1/x , and square root or log… I would have add convolution kernel 3x3 and 5x5
to be able to do blurring, contrast, edge detection or punch reinforcement … the idea
of going back and forth with photoshop or Fiji imageJ …. hum….
There's only one thing I dont understand: Why there is no Selection Tool. For
exemple if I want only save from 2sec to 4sec how can I do it…
Best regards,
qcmv22
. The relationship is almost linear, and for some reason it seems that I'm the only developer who finds it best, so yeah, that's part of the reason why it works so well.There's already convolution, except instead of tiny kernels (I remember using those in the 1990s...) it convolves entire layers together, so you can already do what you mentioned, except better. I'm not sure what a selection tool would do.
