LaserSoft Imaging

He!

I want to thank Kohler from a very good online service. The service has really been like it should be, but is not in every institution. I also want to thank Ian Lyons from the instructive tutorials.

One comment for the one of "Basic steps to optimize high-bit scans". Does one actually really have a need for different workflows for the HDR and for Photoshop. I figure that one can always do the things accoding to the HDR workflow and convert the scanner profile into Adobe RGB in Photoshop. This gives you more flexibility in deciding what to do with raw files. For example one might try adjusting the image in Photoshop and have the same raw file now tailored suitable for editing in HDR.

Then one question I have pondered an answer but been always too lazy to find it out. What actually is Q (=quality) factor. How does adjusting it affect the result.

On 2001-12-05 13:04, klkapi wrote:
He!

I also want to thank Ian Lyons from the instructive tutorials.

Thanks

One comment for the one of "Basic steps to optimize high-bit scans". Does one actually really have a need for different workflows for the HDR and for Photoshop. I figure that one can always do the things accoding to the HDR workflow and convert the scanner profile into Adobe RGB in Photoshop.

Without wishing to cause offence some folk have extreme difficulty grasping that fact. Doing it the easy way in HDR ensures that can't forget or ignore it. You would NOT believe the number of emails I get from folk screaming it looks different in PShop than it did in HDR

This gives you more flexibility in deciding what to do with raw files. For example one might try adjusting the image in Photoshop and have the same raw file now tailored suitable for editing in HDR.

You can leave your raw files untouched, no profile, etc. without an effort.

Then one question I have pondered an answer but been always too lazy to find it out. What actually is Q (=quality) factor. How does adjusting it affect the result.

It relates to the multiplier for the resolution. It is mainly for folk doing CYMK scans for particular screen resolutions. If you set the scanner res at 2000 ppi and the Q factor = 1.5 the scan will be output at 1.5*2000 = 3000ppi

<font size=-1>[ This Message was edited by: ilyons on 2001-12-05 20:15 ]</font>

Saying that "The Q-factor is the quality factor for an image," as the
SF Ai Manual does on p. 29, is but one instance where the Manual
attempts to explain an entry either by using another murky expression,
or by verbose repetition of what the entry says concisely.

It doesn't help either that the same thing is in various places
referred to by various names and in so many different ways. For
instance, in examples for the Scan resolution formula (pp. 151-152),
the "scale-factor" of 1, respectively 0.5, is made to correspond to "a
1:1 scaling-factor," respectively "a scale-factor of 50%"; of course,
1 and 0.5 are factors (i.e. multipliers), but 1:1 and 50% are fractions (i.e. ratios).

<font size=-1>[ This Message was edited by: Tomaz Klinc on 2001-12-06 12:06 ]</font>

Thanks for ilyons for the explanation of the Q factor. I agree with you on your comment to my comment.To leave the raw file without a profile when scanning it with Silverfast A1 is still a third option. The one just has to embed a scanner profile in the HDR or in the Pshop. Still I agree with you. What you suggest in the tutorial is the safe way in getting the correct result.

Another question for you (or for anybody). So far I have used Polaroid as my slide scanner. Have now purchased Nikon Coolscan 4000. A real good thing. However, I have not been able to get the SF-200 slide feeder to work correctly with SilverFast 5.5 and its incorporated job manager. It not a big issue because you can always scan 48 bit raw files with for example NikonScan or with even v**s**n. I am just curious whether the combination should work. If not there is no use in putting effort on it. Any experience on the subject?

dear klkalpi,
The JobManager does not go togehter with a "once-through" feeder like the SF-200.

Just a bit of history:

"Q-Factor" comes from the CMYK press world. It's used to denote how much quality "Q" is lost when outputting to press. For instance, the reason that most Printers want 300dpi files is because they use a Q-Factor of 2, meaning that their presses output 150lpi, but to get the best results, you need to use a 300 dpi file. That extra resolution is lost because of "Factors" such as the dot-gain of the paper, quality of the imagesetter, etc.

I think the other Ian explained it from there, as to it's use in Silverfast.

-Ian A.

<font size=-1>[ This Message was edited by: ianders1 on 2002-03-26 04:38 ]</font>

I just noticed that in v5.52, the "Screen" setting next to the "Q-Factor" is now in LPI (Lines per Inch) and LPCM (Lines per Centimeter) only. I think that this was a really good idea, and will help avoid futher confusion.

-Ian A.

If you already know the desired output resolution of your image (for example, 4000ppi), isn't the Q-Factor and the screen lpi irrelevent since it is only used to calculate the output resolution for the image?

Another way of asking this question is by some examples.

Example 1:

Q-Factor = 1
Screen lpi = 4000
Thus, the image resolution = 4000dpi or ppi

Example 2:
Q-Factor = 2
Screen lpi = 2000
Thus, the image resolution = 4000 dpi or ppi

Doesn't the output image only contain the final resolution and the Q-Factor and Screen lpi information is not contained in the output image file?

(Just trying to understand this Q-Factor thing).

Thanks

Q-Factor History and Relevance explained

Q-Factor indeed comes from tradtional drum scanner technology. When programs in old analog drum scanners were still hard-wired and the input and output were tied together on one axis, a screening dot was done with 2 revolutions of the drum.

Analog Drum Scanner's program
When the scanner operator was selecting an output screen of 152 lpi, the input resolution automatically jumped to 304 dpi (a Q-factor of 2 ).
Later on when the digital drum scanners were introduced, the same setting was kept, although the relation between input and output could be changed and after some discussion that we have had with Hell development and some articles that I have published in different desktop magazines, Hell introduced the so called "High-Speed-Scan".

The Sampling Theorem
The High-Speed scan was using a resolution, which has been defined by the laws of the Sampling Theorem, where the loss in digital sampling would be compensated by the square root of 2 = 1.44 (roughly 1.5)
It took quite a while until many prepress operators realized that the high-speed scan was good-enough for more than 95% of the work and the exception was only with rare high contrast images which were close to line art.

?

? Advantage of adequate resolution
Now if you lower the scan resolution by one quarter from 304 dpi to 225 dpi, (which is what you would roughly get with a Q-factor of 1.5) the file size goes down from 32 MB to 18 MB which is almost half the file size. That means you would have half the memory requirement and dramatically faster processing.

SilverFast introduced Quality Factor in 1994
The first version of SilverFast already introduced the concept of relating the input resolution properly to the output. Which still today is not a standard with most other scan programs, was there right from the start in SilverFast and was easy and safe to use!

best regards

Karl-Heinz Zahorsky
President & Founder
LaserSoft Imaging

I appreciated the explanation and history.

Question. How does interpolated resolution affect the quality of the output?

Personally, I am scanning my photos for the sake of archiving and on-demand printing in the future. I'm using iPhoto to catalog the photos.

Rather than use interpolated resolutions, I am only using hardware resolutions; normally 1000dpi for 35mm film and 1333dpi for APS film; which seem to be high enough to print the photos at A4 reasonable well on today's inkjet printers.

If I was scanning for a specific article with a known output line screen and size, I'd probably use the Q-Factor and High-Speed-Scan settings that SF calculates.

regards
Gregory

Interpolation and Output Quality

Since interpolation cannot generate new information, an image that will be interpolated up, will inevitably loose sharpness. If the method of interpolation is good and will not introduce any artifacts, the loss of sharpness can somewhat be compensated by unsharp masking. Interpolating up for more than 20% does not seem to be reasonable.

In SilverFast for the best quality you should select (under Option --> General) under Interpolation: Antialiased. SilverFast Ai, when using the scaling settings, will always go to the next possible hardware resolution of the scanner and from there interpolate down accordingly.

In SilverFast you can easily check which hardware resolution your scanner would use with the current setting, by pressing the Control key, and for the resultant interpolation down, press Shift in addition.

Interpolating down is mostly without loosing quality. So the best concept is to follow the SilverFast workflow: 1. Set your Q-factor (e.g 1.5) 2. Set your output line screen 3. Set your ouput size. This procedure would assure the best possible quality!

SilverFast doesn't actually execute our scanning/processing instructions until we press the Scan button. Why is it then that we need to follow a recommended workflow when working with the prescans? Wouldn't SilverFast follow its own inbuilt optimised workflow when we press Scan?

regards
Gregory

Workflow in SilverFast?

Dear Gregory,

Basically you are totally free to do anything in SilverFast until you hit the "Scan" button. The only thing that is critical with refernce to scan resolution is Unsharp Masking. The effect of the settings of the USM as monitored inside its preview dialog, is dependent on the scaling parameter ( the final scan resolution). So if you would adjust your USM and afterwards change the scaling or scan resolution, e.g. to higher values, your USM effect would be much weaker and you would have to readjust the settings.
Something similar in a different way would happen if you would set your colour corrections and then click the auto-adjust and then change midtone. Your colour correction might have to be readjusted, since highlight,shadow and midtone might change the whole colour correction effect.
That is why the ScanPilot has (exemplifies) the best possible workflow.

regards
Karl-Heinz Zahorsky

President_LSI is quite off the mark by saying, in an above post, that the square root of 2 is 1.44. In fact this most famous of irrational numbers (i.e., numbers that are not ratios of integers), known as such since the time of Pythagoras, Eudoxus, Plato and Euclid more than 23 centuries ago, and as any high school student and, for that matter photographer, should know, is about 1.41, or 1.414213562373095 with accuracy to the 15-th decimal place. Just hope that SilverFast got it right.

Tomaz.

Be sure, that we always use the formula, instead of a fixed constant, so that the square root of 2 is what it should be .

Eric.