Reflecta DigitDia batch interface: Bugs, annoyances + improv
PostPosted: Sun Jan 08, 2006 5:25 pm
Dear forum users, dear LSI Support,
--- Warning, this is a long post and expresses some frustration. ---
I use the DigitDia 4000 now for about 8 slide magazines and I am
getting more and more annoyed about the interface and the interaction
between SF and the scanner.
I admit that good ideas are there, but the "finetuning" seems to have
been omitted
I really wonder which real tests were done to the use of the Batch
interface for the Reflecta DigitDia slide scanner.
With real tests I mean:
- using the real hardware (which Firmware did YOU use?)
- and actually scanning more than one sildeset in a real session
I have to note the following Bugs, Quirks and Annoyances with SilverFast
Batch Slide Scaning Interface (surely not complete, I still "learn"):
1.) There are two identically looking buttons, that describe the
slide-magazine properties and the position of the slide batch.
- one in the main preview window and
- one in the slide overview window
I observed the following inconsistent behaviour:
a.) when you update data in the dialogue from the main window
not all data is transferred to the button in the slide preview window.
Example: change the setting of the sildes to be scanned and see
if it is visible on "the other side".
-> Why have two times this dialogue?
-> Why can you change the total number of slides only from the main
window's dialogue?
-> Why is the button for the slide overview so far away from the other
two items (button for the slide settings + slide number), having
IT8 calibration and iSRD settings inbetween?
b.) The Window GUI rules are not followed in this (and other)
dialogues. The TAB key lets you wildly jump through the options
and not as expected from top to bottom! Worse, it even changes
the selection boxes by tab-ing through (this should happen with the
cursor keys and TAB only selecting the option-group).
c.) You cannot select the preview slides with the windows-GUI rules
like:
- ctrl-A for selection of ALL slides (this should be limited to
previously definde number of slides in the magazine!)
- you cannot use the page-up and page down to scroll the slides,
nor can you use the cursor keys.
d.) You cannot permanantly change the size of the slide overview window.
The next time you start SF it is again at the old position + size.
2.) When you actually WORK with the scanner you will very fast find out
that it is MUCH more convenient to move the magazine by hand forward from
position 50 to 1 after the slide overview was created.
You update the number, put the magazine at position 1 manually and start
scanning. This really works well until the following point:
- generate a 50 slide magazine preview
- reset to 1
- move magazine
- scan the whole magazine,
- take the next magazine
- reset to 1
- start the overview scan
! here it will hang after the 3rd slide.
It will generate a black mini-slide and will not continue.
If you exit the overview dialogue and get back in, it will only show you 3
slide frames and not the usual 100. When you access the slide-set options
it will still show what you had set (e.g. 50), but you still see only those
3 and no more. There is NO WAY to get out of that except to restart SF.
Q:
- Does the scanner itself "know" about its position in the silde-magazine?
- Do you reset it there when you store the new value (back from 50 to 1)?
It might be a counter overflow inside the scanner?
3.) The Batch-Scan SAVE dialogue (where you construct the filenames) will
forget the path (luckily NOT the filenames) when you kill the SF-Launcher
process. I have to do this from time to time when SF hangs after scanning a
slide (see next bug). Settings should be stored so that this can be restored
like the filenames are.
3a.) The same interface does not comply to Windows GUI rules as the one
mentioned above: TAB makes will never let you step from one part of the
filename to the next, in fact TAB completely excludes the filename section.
When fixing TAB-ing, please ommit the (+) (-) from the sequence.
4.) Occasionally (sorry, NO CLUE) SF hangs after the RAW data were nearly(!?)
transferred and will hook the CPU forever (until you kill it).
I watched that the RAW file grows and grows until about 0.5 MB are missing
from the expected 44-45 MB. I compared an unsuccesfull case with a successful
one for the same slide and the same settings.
Once the problem occurs, it is VERY sticky, i.e restarting SF, switching off
the scanner + on again, nothing helps! Then after 4 retries it worked again.
Size faulty, 4 times identical: 46 224 000 bytes
Size OK same picture, same settings: 46 368 432 bytes
I had to suspend the SF launcher (with sysinternals process explorer) to
be able to copy the file. I also have the associated infrared raw data for
the successful case.
I can supply you with several .raw files but you need to tell me where
I can upload them.
The problem appears with the scanner running on FW 1.02 or 1.04 (no test
with 1.01 or 1.05 (where the slideset preview does not work at all).
It happens with HIREPP ON or OFF.
I use the firewire interface or the USB, it happens on BOTH!
In any case, SF should not hang on incomplete data, but report an error
and stop processing.
Q:
Why are the RAW data only 45 MB? It should be the double size if the scanner
works with 16Bit/color internally. I suspect that also the option 48->24 Bit
reads 48 bits from the device then processes the data and then finally
reduces to 24 bit. Is that not true? If not, why offer to store 48 Bit?
5.) The "delete" key does not work in most text fields. You have to use the
Backspace instead.
6.) When you interrupt a batch scan (CTRL .) then the batch settings are
not updated to the point until where the batch was processed. Only the
actual position in the batch is stored.
You could supply a button "interrupt" batch if the "CTRL ." gets out of the
work in unusual way.
7.) Whoever did the batch scanning surely wants to leave the device
unattended until the time it has done the job.
Please supply a batch progress bar and an information when (date + time)
the job will be finished for the two items:
- generate slide overview
- batch scan
Please recalculate after each slide.
8.) SF cannot sit silently minmized in the system tray when doing the
batch scanning. Dialogue windows pop up again for each slide. If you
prevent them to steal your Foreground Window's focus (windows option)
you can survive, but this is not a nice behaviour.
9.) SF does not update its Preview window when doing a batch scan and you
put it in backround. You can enforce it by bringing the preview window into
focus, but then you get the unprocessed (no light corrections) picture.
It also does not update the slide number in the preview window.
10.) SF windows and dialogues are not refreshed when sending them to
Background and getting them back to foreground until the current scan
operation has completed (this can take some minutes).
11.) When scanning a batch, please supply BOTH numbers,
- the sequence number of the current batch (this is supplied)
- and also the current slide number in the slide-set.
Both may be different. e.g. when you scan a bunch from slide 12-25 where
the sequence number will be 1-13.
12.) Further improvements for the batch slide interface:
a.) Supply an option that the slide magazine is returned to position "1"
automatically after the slide overview was created. This will minimize
the seduction to do that manually to save time when you want to start the
actual batch scan after the overview scan was completed.
b.) Supply a button to reset the position to "1" without moving the magazine
to surely set al variables inside SF and the scanner so that a flawless
operation is ensured. The agazine will be set manually then.
c.) Supply an option "automatically detect empty slides". This will allow
to previewscan a complete magazine without manually setting the ranges.
c1.) Please also supply an option to set a number of consecutive empty
slides after which the rest of the mazine will be skipped.
d.) Suppply a button inside the overview window to start the actual
scan process (go directly there).
13.) Just noted, that the speed the scanner pre-scans the slide for
the "Bildautomatik" depends on the size of the preview window.
I wonder if this is a good idea to have it depending on the size.
I even more wonder why you need that as an extra scan anyhow, since you
cannot (I think) influence the lighting conditions of the scanner, i.e.
exposure time or light strength.
So far for this weekend, quite an impressive list, isn't it?
I will not have accesss to my scanner through the week, so no more
reports to expect for a few days
I know that some bugs may be attributed to the scanner itself, like the
most annoying HANG of SF with incomplete RAW data, but the vast majority
simply looks like bad interface design to me.
I also know that some items are "only" annoyances and improvement proposals,
nevertheless I am a little dissappointed
On your PLUS is definitely the many options you have in processing the
picture data once you have it, but until you get there has much room
for improvement!
This is especially surprising to me, since the DigitDia is supported
since quite some time (previously 3600 without Infrared). Have you never
ever received complaints like that? Are you proud of the interface as it is?
I must say that I am in the Software business myself (not for end users
though) and I can assure you that a batch interface like this one would
not pass my quality check: Release for commercial use DENIED, beta quality!
Thanks for your patience to read so far!
I hope that LSI will show that they take the interface design as serious
as they take the image processing.
bye
Tobias
P.S.: Maybe some DigitDia users can comment on my opinion?
--- Warning, this is a long post and expresses some frustration. ---
I use the DigitDia 4000 now for about 8 slide magazines and I am
getting more and more annoyed about the interface and the interaction
between SF and the scanner.
I admit that good ideas are there, but the "finetuning" seems to have
been omitted
I really wonder which real tests were done to the use of the Batch
interface for the Reflecta DigitDia slide scanner.
With real tests I mean:
- using the real hardware (which Firmware did YOU use?)
- and actually scanning more than one sildeset in a real session
I have to note the following Bugs, Quirks and Annoyances with SilverFast
Batch Slide Scaning Interface (surely not complete, I still "learn"):
1.) There are two identically looking buttons, that describe the
slide-magazine properties and the position of the slide batch.
- one in the main preview window and
- one in the slide overview window
I observed the following inconsistent behaviour:
a.) when you update data in the dialogue from the main window
not all data is transferred to the button in the slide preview window.
Example: change the setting of the sildes to be scanned and see
if it is visible on "the other side".
-> Why have two times this dialogue?
-> Why can you change the total number of slides only from the main
window's dialogue?
-> Why is the button for the slide overview so far away from the other
two items (button for the slide settings + slide number), having
IT8 calibration and iSRD settings inbetween?
b.) The Window GUI rules are not followed in this (and other)
dialogues. The TAB key lets you wildly jump through the options
and not as expected from top to bottom! Worse, it even changes
the selection boxes by tab-ing through (this should happen with the
cursor keys and TAB only selecting the option-group).
c.) You cannot select the preview slides with the windows-GUI rules
like:
- ctrl-A for selection of ALL slides (this should be limited to
previously definde number of slides in the magazine!)
- you cannot use the page-up and page down to scroll the slides,
nor can you use the cursor keys.
d.) You cannot permanantly change the size of the slide overview window.
The next time you start SF it is again at the old position + size.
2.) When you actually WORK with the scanner you will very fast find out
that it is MUCH more convenient to move the magazine by hand forward from
position 50 to 1 after the slide overview was created.
You update the number, put the magazine at position 1 manually and start
scanning. This really works well until the following point:
- generate a 50 slide magazine preview
- reset to 1
- move magazine
- scan the whole magazine,
- take the next magazine
- reset to 1
- start the overview scan
! here it will hang after the 3rd slide.
It will generate a black mini-slide and will not continue.
If you exit the overview dialogue and get back in, it will only show you 3
slide frames and not the usual 100. When you access the slide-set options
it will still show what you had set (e.g. 50), but you still see only those
3 and no more. There is NO WAY to get out of that except to restart SF.
Q:
- Does the scanner itself "know" about its position in the silde-magazine?
- Do you reset it there when you store the new value (back from 50 to 1)?
It might be a counter overflow inside the scanner?
3.) The Batch-Scan SAVE dialogue (where you construct the filenames) will
forget the path (luckily NOT the filenames) when you kill the SF-Launcher
process. I have to do this from time to time when SF hangs after scanning a
slide (see next bug). Settings should be stored so that this can be restored
like the filenames are.
3a.) The same interface does not comply to Windows GUI rules as the one
mentioned above: TAB makes will never let you step from one part of the
filename to the next, in fact TAB completely excludes the filename section.
When fixing TAB-ing, please ommit the (+) (-) from the sequence.
4.) Occasionally (sorry, NO CLUE) SF hangs after the RAW data were nearly(!?)
transferred and will hook the CPU forever (until you kill it).
I watched that the RAW file grows and grows until about 0.5 MB are missing
from the expected 44-45 MB. I compared an unsuccesfull case with a successful
one for the same slide and the same settings.
Once the problem occurs, it is VERY sticky, i.e restarting SF, switching off
the scanner + on again, nothing helps! Then after 4 retries it worked again.
Size faulty, 4 times identical: 46 224 000 bytes
Size OK same picture, same settings: 46 368 432 bytes
I had to suspend the SF launcher (with sysinternals process explorer) to
be able to copy the file. I also have the associated infrared raw data for
the successful case.
I can supply you with several .raw files but you need to tell me where
I can upload them.
The problem appears with the scanner running on FW 1.02 or 1.04 (no test
with 1.01 or 1.05 (where the slideset preview does not work at all).
It happens with HIREPP ON or OFF.
I use the firewire interface or the USB, it happens on BOTH!
In any case, SF should not hang on incomplete data, but report an error
and stop processing.
Q:
Why are the RAW data only 45 MB? It should be the double size if the scanner
works with 16Bit/color internally. I suspect that also the option 48->24 Bit
reads 48 bits from the device then processes the data and then finally
reduces to 24 bit. Is that not true? If not, why offer to store 48 Bit?
5.) The "delete" key does not work in most text fields. You have to use the
Backspace instead.
6.) When you interrupt a batch scan (CTRL .) then the batch settings are
not updated to the point until where the batch was processed. Only the
actual position in the batch is stored.
You could supply a button "interrupt" batch if the "CTRL ." gets out of the
work in unusual way.
7.) Whoever did the batch scanning surely wants to leave the device
unattended until the time it has done the job.
Please supply a batch progress bar and an information when (date + time)
the job will be finished for the two items:
- generate slide overview
- batch scan
Please recalculate after each slide.
8.) SF cannot sit silently minmized in the system tray when doing the
batch scanning. Dialogue windows pop up again for each slide. If you
prevent them to steal your Foreground Window's focus (windows option)
you can survive, but this is not a nice behaviour.
9.) SF does not update its Preview window when doing a batch scan and you
put it in backround. You can enforce it by bringing the preview window into
focus, but then you get the unprocessed (no light corrections) picture.
It also does not update the slide number in the preview window.
10.) SF windows and dialogues are not refreshed when sending them to
Background and getting them back to foreground until the current scan
operation has completed (this can take some minutes).
11.) When scanning a batch, please supply BOTH numbers,
- the sequence number of the current batch (this is supplied)
- and also the current slide number in the slide-set.
Both may be different. e.g. when you scan a bunch from slide 12-25 where
the sequence number will be 1-13.
12.) Further improvements for the batch slide interface:
a.) Supply an option that the slide magazine is returned to position "1"
automatically after the slide overview was created. This will minimize
the seduction to do that manually to save time when you want to start the
actual batch scan after the overview scan was completed.
b.) Supply a button to reset the position to "1" without moving the magazine
to surely set al variables inside SF and the scanner so that a flawless
operation is ensured. The agazine will be set manually then.
c.) Supply an option "automatically detect empty slides". This will allow
to previewscan a complete magazine without manually setting the ranges.
c1.) Please also supply an option to set a number of consecutive empty
slides after which the rest of the mazine will be skipped.
d.) Suppply a button inside the overview window to start the actual
scan process (go directly there).
13.) Just noted, that the speed the scanner pre-scans the slide for
the "Bildautomatik" depends on the size of the preview window.
I wonder if this is a good idea to have it depending on the size.
I even more wonder why you need that as an extra scan anyhow, since you
cannot (I think) influence the lighting conditions of the scanner, i.e.
exposure time or light strength.
So far for this weekend, quite an impressive list, isn't it?
I will not have accesss to my scanner through the week, so no more
reports to expect for a few days
I know that some bugs may be attributed to the scanner itself, like the
most annoying HANG of SF with incomplete RAW data, but the vast majority
simply looks like bad interface design to me.
I also know that some items are "only" annoyances and improvement proposals,
nevertheless I am a little dissappointed
On your PLUS is definitely the many options you have in processing the
picture data once you have it, but until you get there has much room
for improvement!
This is especially surprising to me, since the DigitDia is supported
since quite some time (previously 3600 without Infrared). Have you never
ever received complaints like that? Are you proud of the interface as it is?
I must say that I am in the Software business myself (not for end users
though) and I can assure you that a batch interface like this one would
not pass my quality check: Release for commercial use DENIED, beta quality!
Thanks for your patience to read so far!
I hope that LSI will show that they take the interface design as serious
as they take the image processing.
bye
Tobias
P.S.: Maybe some DigitDia users can comment on my opinion?