Wavelet-compressed HRV data


David J Taylor GM8ARV 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🇪🇺
 

But we don't know if these lossless wavelet compressed HRV's are
here to stay. (I really hope so !)

Arne, Eumetsat are looking for any reports of problems with the new
format, so if they don't get any such reports I think the likelihood
of keeping the format increases. If anyone has problems with the
new HRV format please do speak now!!

So I should do a new comparison on lossy compressed HRV's from IMPF
(29 Oct) saved in jpg86 and png.
It would be great to get a good recommendation for the JPEG save
quality I set as default. Of course, the user can alter this to get
either smaller files or a better quality, and it is a subjective
measurement....

Currently I am determining new min/max values on MSG-1 and GOES
images using the 10 bit Rob Alblas XRIT2PIC decoder, will let all
know my results later this week.
First observation: The IMPF system does use the 0..1023 range much
better compared to the IIPS.
The min values should be unchanged, as all MSG-1 channels are
processed so that zero radiance is count 51. That was true of the
IQGSE software and remains true with the IMPF software.

You can set a channel temporarily to min:0, max:1023, gamma: 100,
thermal cal:no to see exactly what range is transmitted. The MSG-1
LRIT channels are already processed with those settings, by the
way.

Given the resulting image histogram then says 13..200 (say), the
corresponding values to set for main and max would be four times
that, 52..800, say. There is no one "correct" setting for gamma.
The value of 140 (gamma = 1.4) is designed so that single count
sensor changes around black level appear as single count image pixel
changes, thus preserving as much of the dynamic range as possible in
the dark parts of the image.

Also beware that when measuring the histogram, the off-earth values
are set to zero rather than the minimum sensor count (this
translates to a value of 1 in the JPEG saved LRIT images I just
checked), so be sure to ignore any histogram values below 13 in
unprocessed images.

Cheers,
David


a_van_belle
 

--- In MSG-1@yahoogroups.com, "David Taylor" <david-taylor@b...>
wrote:
That's a good point, Ian. I have increased the default JPEG save
level from 82 to 86, but I believe that Arne van Belle wants to
update some tests on the save level setting. It will be
interesting to see his new results.

Hello David and others,

Did some tests, comparing the 28 October IIPS lossy HRV to 30
October's IMPF losless HRV. This showed a drastic improvement on
details and gave smooth grey gradients. All "jpeg artefacts" are gone
when saving losless HRV to PNG.
But we don't know if these lossless wavelet compressed HRV's are here
to stay. (I really hope so !)

So I should do a new comparison on lossy compressed HRV's from IMPF
(29 Oct) saved in jpg86 and png.

Currently I am determining new min/max values on MSG-1 and GOES
images using the 10 bit Rob Alblas XRIT2PIC decoder, will let all
know my results later this week.
First observation: The IMPF system does use the 0..1023 range much
better compared to the IIPS.

Greetings,
Arne van Belle


Ian S Deans <ian@...>
 

Further to my previous mail, I have now tried saving channel 12 with a jpg
setting of 100 and that has appeared to get rid of any artefacts. However
the file size has increased from just over 8mb to over 22mb. I then saved
channel 12 on the next cycle as png and the file size was also approx. 22mb.
Correct me if I am wrong, but it would seem to me that there is no benefit
in using jpg 100, better with the lossless png.

Ian.


Ian S Deans <ian@...>
 

Sorry about the "typo" in my previous mail. Should have been "exceeded my
best expectations"

Ian.


Ian S Deans <ian@...>
 

Guy ( and David ) thanks for your reply.

The only reason I try to keep HRV image size reasonable is that I do use
GSS4 quite a lot when everything else is still running. Although my Athlon
1200 with 768mb ram has succeeded my best expectations, it is being
seriously pushed. I will try a jpg setting of 100 ( when MSG starts up again
!! ) although I suspect at that level the size of file will be close to png
anyway.

Thanks again

Regards
Ian.

----- Original Message -----
From: "Guy Martin" <agm@tonbridge-school.org>
To: <MSG-1@yahoogroups.com>
Sent: Monday, November 03, 2003 8:18 AM
Subject: Re: [MSG-1] Wavelet-compressed HRV data


Hi Ian,

I'm saving them as jpg with a setting of 100. Saving does now take longer
but the images are now excellent. I'm using an Athlon 1200 with 512Gb and
the processor spends time at 100% so an upgrade is planned in the future as
it is my main working machine. I'm not concerned about image size as I only
archive interesting features and will copy them to CD.

Cheers, Guy


Guy Martin <agm@...>
 

Hi Ian,

I'm saving them as jpg with a setting of 100. Saving does now take longer but the images are now excellent. I'm using an Athlon 1200 with 512Gb and the processor spends time at 100% so an upgrade is planned in the future as it is my main working machine. I'm not concerned about image size as I only archive interesting features and will copy them to CD.

Cheers, Guy

----- Original Message -----
From: Ian S Deans
To: MSG-1@yahoogroups.com
Sent: Sunday, November 02, 2003 9:10 PM
Subject: Re: [MSG-1] Wavelet-compressed HRV data


Guy,

Are you saving the HRV images as png or jpg. If jpg what setting are you
using. I have increased the jpg setting to 90, but I am still getting
artefacts, although slight improvement. Most noticeable as you say as the
terminator approaches. I am running everything on one Athlon 1200
extremely successfully, but am endeavouring to keep HRV image size
to a reasonable level without too much artefacts/pixellation.


Ian.

----- Original Message -----
From: "Guy Martin" <agm@tonbridge-school.org>
To: <MSG-1@yahoogroups.com>
Sent: Sunday, November 02, 2003 10:37 AM
Subject: Re: [MSG-1] Wavelet-compressed HRV data


> It has certainly improved picture quality. Animating the terminator now
shows a smooth transition to black with no artifacts or pixellation. The
system appears to cope just as well.
>



Yahoo! Groups Sponsor
ADVERTISEMENT




To unsubscribe from this group, send an email to:
MSG-1-unsubscribe@yahoogroups.com



Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.

----------

http://www.gordano.com - Messaging for educators.


David J Taylor GM8ARV 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🇪🇺
 

--- In MSG-1@yahoogroups.com, "Ian S Deans" <ian@i...> wrote:
Guy,

Are you saving the HRV images as png or jpg. If jpg what setting
are you using. I have increased the jpg setting to 90, but I am
still getting artefacts, although slight improvement. Most
noticeable as you say as the terminator approaches. I am running
everything on one Athlon 1200 extremely successfully, but am
endeavouring to keep HRV image size to a reasonable level without
too much artefacts/pixellation.


Ian.
That's a good point, Ian. I have increased the default JPEG save
level from 82 to 86, but I believe that Arne van Belle wants to
update some tests on the save level setting. It will be interesting
to see his new results. To convert the 10-bit data to an 8-bit
display format compromises are required, although I believe that my
default settings of min:51, max:1023 and gamma:140 (i.e. 1.40) are
close to the optimum. Of course, you can play with the settings to
suit your needs. The GMS (GOES-9) images are set (at source) to a
gamma of 2, and look rather washed out to me....

For my own comparison tests, I saved in PNG and not JPEG format.

Cheers,
David


Ian S Deans <ian@...>
 

Guy,

Are you saving the HRV images as png or jpg. If jpg what setting are you
using. I have increased the jpg setting to 90, but I am still getting
artefacts, although slight improvement. Most noticeable as you say as the
terminator approaches. I am running everything on one Athlon 1200
extremely successfully, but am endeavouring to keep HRV image size
to a reasonable level without too much artefacts/pixellation.


Ian.

----- Original Message -----
From: "Guy Martin" <agm@tonbridge-school.org>
To: <MSG-1@yahoogroups.com>
Sent: Sunday, November 02, 2003 10:37 AM
Subject: Re: [MSG-1] Wavelet-compressed HRV data


It has certainly improved picture quality. Animating the terminator now
shows a smooth transition to black with no artifacts or pixellation. The
system appears to cope just as well.


Guy Martin <agm@...>
 

It has certainly improved picture quality. Animating the terminator now shows a smooth transition to black with no artifacts or pixellation. The system appears to cope just as well.

Regards, Guy

----- Original Message -----
From: David Taylor
To: MSG-1@yahoogroups.com
Sent: Sunday, November 02, 2003 9:05 AM
Subject: [MSG-1] Wavelet-compressed HRV data


I see we are still on wavelet-compressed HRV data. I wonder if this
will be a permanent change?
Yahoo! Groups Sponsor
ADVERTISEMENT




To unsubscribe from this group, send an email to:
MSG-1-unsubscribe@yahoogroups.com



Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.

----------

http://www.gordano.com - Messaging for educators.


David J Taylor GM8ARV 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🇪🇺
 

I see we are still on wavelet-compressed HRV data. I wonder if this
will be a permanent change? The larger file size shows up well on
the bandwidth graphs - look at the increase from last Wednesday
lunchtime! I haven't seen any ill effects here such as an increase
in missing segments, file sequence errors, or the PC simply failing
to cope with the greater data rate.

http://www.david-taylor.myby.co.uk/mrtg/hermes_dvb.html

I wonder if this would have been possible with direct dissemination,
rather than the greater data rates used on the DVB service?

Cheers,
David