I have two images, captured from a netcam.
There are no "real" changes between the images:
1. There is a slight change in reflection from the sun
2. A plant's leaves move a little bit because of the wind
Perceptually, there is no change.
How can PerceptualDiff ignore those changes?
I can send/upload the images somewhere for you to see if you are around...
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I have two images, captured from a netcam.
There are no "real" changes between the images:
1. There is a slight change in reflection from the sun
2. A plant's leaves move a little bit because of the wind
Perceptually, there is no change.
How can PerceptualDiff ignore those changes?
I can send/upload the images somewhere for you to see if you are around...
I tried values all over the fov range, and I am seeing no changes from 0.1 to 1.2, and from 1.3 on i see lots of changes. So I din't see a value that works for me yet. How would you set the configuration to compare 2 antialiased images, where pixels may move a bit from one image to the next?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I would downsample by the number of acceptable pixel movements before comparing.
It was designed more for dealing with shader unit testing rather than
translations in pixels
where each pixel is more or less in the same spot but due to
monte-carlo sampling may have slightly different RGB values.
If you wanted something that 'looked' similar, maybe either downsample
or Gaussian blur or median filter the images before comparing.
I tried values all over the fov range, and I am seeing no changes from 0.1
to 1.2, and from 1.3 on i see lots of changes. So I din't see a value that
works for me yet. How would you set the configuration to compare 2
antialiased images, where pixels may move a bit from one image to the next?
I'd like to say thanks by offering you a license of any of the software titles that I publish - see ist97.com for a list.
BTW, some of these projects/libraries are already free, and can be downloaded on sf.net ;)
Best regards.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I have two images, captured from a netcam.
There are no "real" changes between the images:
1. There is a slight change in reflection from the sun
2. A plant's leaves move a little bit because of the wind
Perceptually, there is no change.
How can PerceptualDiff ignore those changes?
I can send/upload the images somewhere for you to see if you are around...
Try decreasing the field of view parameter to 10.. that should fix
most small changes
On Tue, Jun 21, 2016 at 5:51 PM, Sal Ingrilli shortpasta@users.sf.net wrote:
--
Yee Yang Li Hector
google.com/+HectorYee
I tried values all over the fov range, and I am seeing no changes from 0.1 to 1.2, and from 1.3 on i see lots of changes. So I din't see a value that works for me yet. How would you set the configuration to compare 2 antialiased images, where pixels may move a bit from one image to the next?
I would downsample by the number of acceptable pixel movements before comparing.
It was designed more for dealing with shader unit testing rather than
translations in pixels
where each pixel is more or less in the same spot but due to
monte-carlo sampling may have slightly different RGB values.
If you wanted something that 'looked' similar, maybe either downsample
or Gaussian blur or median filter the images before comparing.
On Thu, Jun 23, 2016 at 11:44 AM, Sal Ingrilli shortpasta@users.sf.net wrote:
--
Yee Yang Li Hector
google.com/+HectorYee
Great advice - thank you for what you have done!
Most welcome ... I'm sometimes surprised that such an obscure tool
finds unexpected uses!
On Fri, Jun 24, 2016 at 4:34 PM, Sal Ingrilli shortpasta@users.sf.net wrote:
--
Yee Yang Li Hector
google.com/+HectorYee
I'd like to say thanks by offering you a license of any of the software titles that I publish - see ist97.com for a list.
BTW, some of these projects/libraries are already free, and can be downloaded on sf.net ;)
Best regards.