Filmscanner-Forum

Both lines - the morse code and the continuous - are in the same direction, the scanning direction. Sorry, forgot to rotate the pictures.

Yes, the line is 2 pixels wide. But it's not the only line. I have around 7 or 8 on the whole sensor. 7 or 8 groups of 2 pixels. I'll try the black slide approach as soon as I have some time to investigate.

The morse code is basically 12p pixels long X 2 pixels wide. It's 24 pixels away from the next iteration. 12 pixels over a length of 36 pixels. 1/3.

The Coolscan 8000 and 9000 both have 3 lines CCDs unlike the 5000 that only have 2. There is a mode in both Vuescan and Nikonscan that allows to only use one line in order to get a better picture without banding. It multiplies the time of scanning by a factor of 3 (sounds pretty logical). My theory is basically that the first CCD row on my sensor is riddled with hot pixels. The 2 others lines do not show that flaw.

When using the fine mode, the first row is used and shows a continuous line. Too bad it is the first row that is full of hot pixels... Oh well, it is what it is.

By the way, only Vuescan allows the use of the famous "fine mode" with the Coolscan 5000. It's basically a hack.

That's odd.

So, the problem are twins of 2 pixels on the first CCD row, and also twins of 2 pixels at exactly the same spot on the second row of the CCD. Seems not to be singular hot pixels.

I guess we need Jossie's advice.

Looking at your pictures, the line is in the dark grey area, that is, still with some light.The test with the "black" slide my help further.

Bobby hat geschrieben: ↑Dienstag 18. April 2023, 12:03 he problem are twins of 2 pixels on the first CCD row, and also twins of 2 pixels at exactly the same spot on the second row of the CCD

Not as far as I can see. Two "warm" pixels on one CCD could account for what we see in the image. Also in fine mode the stripe is two pixels wide.

The length of the stripe and of the gap in the morse image reflects the physical separation of the CCDs.

Let's wait to see the dark image, which should clarify if in the end it is not a dust spec producing some weird flatfield remanent feature.

Hermann-Josef

Weird. I posted a link to the black picture that seems to have been deleted.
Another try :

http://www.4-16.com/scanner/black.jpg

From this I indeed would conclude, that this CCD has quite a few warm pixels of varying intensity:


Also the bias (or dark) has structure in it, if the slide was really completely dark.

Hermann-Josef

Your picture can be opened in a browser and shows up as in the screenshot below.
(You can upload picture files when writing a text by using the feature below the text window)

Question: what are the black ribbons above and below the grey area?

There are at least 8 vertical stripes in the grey area as you have described, and these stripes are of varying intensity.

Can you make a 16 bit histogram, or better a 16 bit horizontal profile with pixels at the x-axis?

Sorry, I am puzzled by the scan dimensions. In my answer, I had assumed that the scan direction was vertical?

Hermann-Josef

PS: For analysis with imageJ it would be best to supply the original TIF image, e.g. via filebin.net.

Well I guess I have to hope the CCD I bought is in better shape.

That picture was made with a piece of thick black cardboard. It also shows that the LED backlight is quite uneven. The Red channel is particularly different from the 2 others.

Brightness was pushed 4 or 5 stops in Lightroom to highlight the hot pixels. Now what I find odd is the fact that it always a two pixels wide line.

Stupid question, sorry. What scanner did you use, and what resolution?

There's no stupid question

Coolscan 9000. Full resolution width in medium format, 8964 pixels wide

??

The Coolscan 9000 has a native resolution of 4000 ppi, and a picture format of 6x9 cm. If the scan is made with the CCD at the broad side (9 cm), then we would have 14'173 Pixels in the row. You inform that the width is 8964 pixels, Jossie's graph shows about 9000.

What part of the whole image does your picture show?

The CCD width is around 6cm, the 9cm limit is along the CCD axis. I just scanned 2 or 3cm along the CCD axis, just to show the hot pixels. Didn't seem that interesting to scan a 9cm long black stripe.

oK, that is now clear: you used the native resolution, the picture is at CCD width, and the stripes are on a part of the whole picture (oK so), and go in scan direction. Sorry to ask for such fundamentals, but its better to ask than to draw the wrong conclusions.

It would also be beneficial to know what the thick black ribbons above the grey picture area represent.

Then, as Jossie suggested, it would be good if you could upload the scan, eg at filebin, so that we could analyze the picture in more detail as you could do with Lightroom (it should better be done in ImageJ).