Scanner Comparison 2019 [Epson Flatbed | Eversmart Flatbed | Drum Scanners]

[Pere Casals]

I am sure just as amazing colors can be achieved with some post editing on the flatbed scans.

Of course, the drum has no stray light in the optic system. But stray level can be removed mostly by adjusting the input level in the Ps curve tool, in some images it would require additional work, IMHO.

After removing the stray, if we equalize the images in photoshop we see that images are mostly equivalent, being possible to match one to the other, except (I guess) for the extreme densities.

Regarding the palette, different sensors have different spectral responses being the cross channel different, not better or worse in principle. For an exact color match we may require a 3D LUT.

[Pali K]

You're right Pere. I am sure it could be improved and that is why I am posting RAW files so hopefully others can do some post-edit conclusions as well later. I am scanning through and reviewing the images and though this area of the image shows some interesting differences that are worth sharing.



What is most interesting is that in real life, DPI resolving power may not be adding a whole lot if the image itself doesn't have the resolution to begin with. It will be nice to see how slides look side by side and I expect to see bigger differences there.

[Pere Casals]

To me the EPSON requires a 2.5pix 50% sharpening radius to match a bit, and a little touch in the curves shaping in S.

[Pali K]

Scanview Scanmate 5000 Drumscanner | ColorQuartet 5.2.2 WIN | IT8 Colortarget 8 BIT RAW

LINK TO FILE

[Pali K]

Scanview Scanmate 5000 Drumscanner | ColorQuartet 5.2.2 WIN | IT8 Colortarget 8 BIT RAW

This is the best resolution my Scanmate 5000 can achieve.

LINK TO FILE



If I zoom the image 200%, I can see lines at Group 6 Element 3 which puts the DPI at ~4100 DPI.

[Pali K]

Scanview Scanmate 5000 Drumscanner | ColorQuartet 5.2.2 WIN | Kodak Portra 160 Color Negative 8 BIT RAW

LINK TO FILE



The best ColorPerfect conversion I could achieve without increasing saturation. LINK TO 5000 DPI FILE

[Ted Baker]

I hope no one minds me saying this but as I expected, the colors on the drum are just another level without touching anything. I am sure just as amazing colors can be achieved with some post editing on the flatbed scans.

That is indeed interesting, all of these devices have the WRONG spectral response for color negative. Though i suspect the drum scanners with there more sensitive PMT have been engineered to have a narrower spectral response. This would may make it easier to correct the inherent errors, or not... (depending where the spectrum is...)

AFAIK, colorperfect makes no attempt to fix these problems.

[Pere Casals]

all of these devices have the WRONG spectral response for color negative.

Ted, I don't agree. All those devices are amazingly good for color.

Also all can be IT8 calibrated. We also may design our own calibration to target any spectral hue in the negative palette to our (reversed) RGB palette.

A color scanner makes always a color separation. Light from film has an spectral nature but color dyes on pixels or on PMTs makes the R-G-B color separation.

Theory ruling color separations in the movie industry (for archival) it's the same that rules separations/processing from a scanner. With that knowledge it can be nailed what wanted in color processing. Today this just requires 3D LUTs and a colorist technician. At Disney they have plenty.

Think that from each RGB reading from the scanner we may even map the spectrum that the particular film has from that RGB combination, so its about being good in the processing, we have no limitation from the gear (beyond DMax).

[Pali K]

Heidelberg Tango Drumscanner | Newcolor 2000 MAC | IT8 Colortarget 16 BIT TIFF

Note: Tango has no RAW with my setup but this is with everything (end points, curves, sharpness, and etc.) turned off. However, Tango has an auto white-balance routine that has already corrected any color bias seen by the PMTs.

LINK TO FILE

[Pali K]

Heidelberg Tango Drumscanner | Newcolor 2000 MAC | IT8 Colortarget 16 BIT TIFF

This is the best resolution my Tango can achieve. I think the resolution is between Group 6 Element 4 or 5 putting the DPI resolution at ~4600 - 5100 DPI.

LINK TO FILE