PIE Primefilm XA Plus

[Romanko]

fake mathematical wizardry

Demosaicing

 

[dwstudeman]

Demosaicing

Yes, that is what people call it. It sounds more palatable. I know it sounds like I am being unfair because in reality, until we can get a sensor with three color layers at each pixel, this is what we have and I have seen most versions of demosaicing produce very convincing images and even on a scanned film image, once you adjust color it blows what I said out the window. Maybe control isn't such a bad thing, hmmm.

 

[koraks]

Welcome to Photrio @dwstudeman!

As I am sure everyone knows, part of the beauty of film is that every photosite has all three primaries in layers, and line scanners such as this are NOT Bayer, They have three scanner heads that are in rows of ONE color per head, one red, one blue

There's an interesting exploration of at least two different scanners (Epson V700/750 and Nikon 4000) in this pdf, which gets fairly technical and seems well-substantiated: http://dasch.rc.fas.harvard.edu/papers/Scannerevaluation1.pdf

Note that not all colors are actually collected at the same time for each spot on the film. Due to the distance between the CCD strips, each pixel will consist of several signal inputs acquired successively over multiple longitudinal steps. While this results in pixels that are indeed constructed from R, G and B signals (sometimes more than one, each), the fact that the sampling necessarily occurs in different moments introduces other problems.

There's always a tradeoff, somewhere.

 

[dwstudeman]

Welcome to Photrio @dwstudeman!


There's an interesting exploration of at least two different scanners (Epson V700/750 and Nikon 4000) in this pdf, which gets fairly technical and seems well-substantiated: http://dasch.rc.fas.harvard.edu/papers/Scannerevaluation1.pdf

Note that not all colors are actually collected at the same time for each spot on the film. Due to the distance between the CCD strips, each pixel will consist of several signal inputs acquired successively over multiple longitudinal steps. While this results in pixels that are indeed constructed from R, G and B signals (sometimes more than one, each), the fact that the sampling necessarily occurs in different moments introduces other problems.

There's always a tradeoff, somewhere.

Yes, there seems to always be a tradeoff in everything we design.

 

[Romanko]

until we can get a sensor with three color layers at each pixel

Foveon X3 is one such sensor. Apparently, Sigma is busy prototyping a new incarnation of Foveon.

 

[brbo]

Note that not all colors are actually collected at the same time for each spot on the film. Due to the distance between the CCD strips, each pixel will consist of several signal inputs acquired successively over multiple longitudinal steps. While this results in pixels that are indeed constructed from R, G and B signals (sometimes more than one, each), the fact that the sampling necessarily occurs in different moments introduces other problems.

Not only at different moments, but also different positions. For a "standard" CCD linear sensor (3-line sensor with RGB filter) to get one line with all three R, G, B channels you have to take 3 readings and move film twice between readings. Results from staggered 6-line sensors like those in V700/V800 are obviously subject to even more interpretation. Not exactly "false math", but mechanical and electronic variation/imprecision.

Nikon Coolscan 8000/9000 have monochromatic (no RGB filter) 3-line CCD sensor and you can enable "fine" mode where three R, G, B readings are taken from the same line on the sensor and the film is not moved between readings.

Foveon would indeed be a perfect sensor, but Sigma (with Quattro) actually strayed away (slightly) form the original concept to keep up with CMOS sensors which are apparently perceived as good enough for general photography.

 

[dwstudeman]

Not only at different moments, but also different positions. For a "standard" CCD linear sensor (3-line sensor with RGB filter) to get one line with all three R, G, B channels you have to take 3 readings and move film twice between readings. Results from staggered 6-line sensors like those in V700/V800 are obviously subject to even more interpretation. Not exactly "false math", but mechanical and electronic variation/imprecision.

Nikon Coolscan 8000/9000 have monochromatic (no RGB filter) 3-line CCD sensor and you can enable "fine" mode where three R, G, B readings are taken from the same line on the sensor and the film is not moved between readings.

Foveon would indeed be a perfect sensor, but Sigma (with Quattro) actually strayed away (slightly) form the original concept to keep up with CMOS sensors which are apparently perceived as good enough for general photography.

I was referring to the bayer sensor which the scanners shouldn't have but the word false or fake admittedly is a bit strong. Were it not for Bryce Bayer in 1975 we may not have had color all these years.

The Nikon is one of the few that can scan uncut rolls which is how most of my negatives over the last 30 years are stored. Purportedly the PrimeFilm XA series does well with uncut rolls as far as flatness and focus. Only one way for me to find out. I wish I had the top Minolta scanner that was produced. I just don't like to cut up negative strips though. They are kept in air-tight plastic canisters made for this, no, not like the one the film comes in, too tall and too small a diameter. As far as uncut film, some labs have the Noritsu or the Fuji Frontier which still works very well. I'll often have my negatives developed and then have them run through the Frontier or Noritsu scanner. The only weakness is that the 645 format with less DPI available in Medium Format could stand more DPI on the Frontier and Noritsu but just going 6x7 ends up with a lot of megapixels even at the lower dpi than 35mm.

 

[dwstudeman]

Welcome to Photrio @dwstudeman!


There's an interesting exploration of at least two different scanners (Epson V700/750 and Nikon 4000) in this pdf, which gets fairly technical and seems well-substantiated: http://dasch.rc.fas.harvard.edu/papers/Scannerevaluation1.pdf

This is a good read, thanks!

 

[JohnWolf]

I ordered a PrimeFilm XA a couple weeks ago to compare it to my Plustek 8200. The XA scan quality and speed seem superior to me.

I sent it back today, however, for a couple reasons. The scan surface is covered in lint. I tried blowing it off, but no luck. The other is that negatives won't align properly for scanning. I don't envision scanning whole rolls, but my six-negative strips feed with random offsets so the whole frame never aligns accurately in the scan window.

I use 64-bit Vuescan Professional. Had to install the latest version, but it works fine. I thought the scanner's physical buttons did not work with Vuescan, but Ed Hamrick told me to hold them down longer and that worked.

Overall, I like the scanner better than the Plustek. I returned it to Amazon and ordered a new one from B&H, thinking they likely have more turnover and newer products.

John

 

[Romanko]

Purportedly the PrimeFilm XA series does well with uncut rolls as far as flatness and focus.

It might be doing well with regard to flatness and focus. I was never able to scan a whole roll of film without jamming or crashing. Even a 6-frame strip is often a challenge even with a freshly developed and flat film. Forget scanning decades old negatives stored in rolls. I was disappointed with PrimeFilm XA and regret not sending it back.

 

[John Wolf]

As said above, I returned the first scanner and have since received a second. On this one the autoload function always advances to the second frame, and there's no way to get to the first frame. Vendor support says it's a hardware defect and the scanner should be returned.

I'm giving up on product. Too bad, because I like the results very much.

John

 

[Radost]

Great scanner. Needs Silver fast for best results.
I purchased and returned when I found that it does not work with silverfast

 

[Romanko]

Great scanner. Needs Silver fast for best results.
I purchased and returned when I found that it does not work with silverfast

So, this is a great scanner that requires Silverfast to work properly but you had to return it because it does not work with Silverfast, right?

 

[Radost]

So, this is a great scanner that requires Silverfast to work properly but you had to return it because it does not work with Silverfast, right?

Correct. At that moment it was not. I don’t know if it is now.

 

[Radost]

 

[braxus]

I really dont see any difference with the XAs and the new coming XA Plus. They look identical to me, even in the specs. I wonder if the driver for this new scanner will work on the older XAs? And they havent seemed to solve the issue of getting each successive frame centered on the scan, without readjustment.

 

[TomR55]

Re software: I don't think that the developers at SilverFast would be collecting paychecks to create software that works for this device if only a few minor tweaks were sufficient to host this device. I agree that their form factors appear very similar (if not identical), but I ask why would PIE spend the resources to create a "new" device that offered no measurable advantage over what is already available, supported and used by customers? That said, the required software modifications might be minor compared to the potential return on investment if this scanner offers significant improvements over the previous model(s). (If I have a moment, I'll check the Vuescan website to see if this scanner is supported ....)

Re hardware: I use the XA version of this scanner and have no problems dependably scanning six-frame strips, but that might be dumb luck.

FWIW: I think that a reliable "lab" scanner that played well in the consumer marketplace would be an all-around win for the manufacturers and the software companies that supported this device and, most importantly, for the vast majority of film consumers.

 

[Radost]

Re software: I don't think that the developers at SilverFast would be collecting paychecks to create software that works for this device if only a few minor tweaks were sufficient to host this device. I agree that their form factors appear very similar (if not identical), but I ask why would PIE spend the resources to create a "new" device that offered no measurable advantage over what is already available, supported and used by customers? That said, the required software modifications might be minor compared to the potential return on investment if this scanner offers significant improvements over the previous model(s). (If I have a moment, I'll check the Vuescan website to see if this scanner is supported ....)

Re hardware: I use the XA version of this scanner and have no problems dependably scanning six-frame strips, but that might be dumb luck.

FWIW: I think that a reliable "lab" scanner that played well in the consumer marketplace would be an all-around win for the manufacturers and the software companies that supported this device and, most importantly, for the vast majority of film consumers.

You think wrong. SIlver fast is making it compatible.
Also compatibility with apple silicon and faster speed are not a small things.

 

[TomR55]

You think wrong. SIlver fast is making it compatible.
Also compatibility with apple silicon and faster speed are not a small things.

Thanks for the clarification.

 

[dwstudeman]

I really dont see any difference with the XAs and the new coming XA Plus. They look identical to me, even in the specs. I wonder if the driver for this new scanner will work on the older XAs? And they havent seemed to solve the issue of getting each successive frame centered on the scan, without readjustment.

You won't see a difference in form factor as that external form factor has been around for quite some time. Even the old Braun has the same form factor.

I would be inclined to get the XAs as the XA Plus, while being faster, does not work in Linux at present. I am not sure if Ed will be able to get Vuescan to work with the XA Plus for 35mm. Perhaps someone who is regularly in contact with him has an idea of what is wrong. I had heard that the USB implementation is where the problem is but I have patched USB drivers in the past to solve problems with certain hardware and recompiled the Linux kernel after hacking the code. If it's in the Linux Kernel USB stack where the problem is, that can certainly be modified and even set to only pass the new parameters when that scanner is plugged in and UDEV identifies the hex manufacturer ID and Device ID.

 

[JohnWolf]

I had two XA Plus recently and they worked with Vuescan (Windows). They had other problems, but Vuescan compatibility was good.

John

 

[Radost]

I had two XA Plus recently and they worked with Vuescan (Windows). They had other problems, but Vuescan compatibility was good.

John

Silverfast gives me a lot better results with XAS.

 

[dwstudeman]

I had two XA Plus recently and they worked with Vuescan (Windows). They had other problems, but Vuescan compatibility was good.

John

Yes, with the Cyberview driver, you can in Windows.

 

[dwstudeman]

Silverfast gives me a lot better results the XAS.

The XAS is a different machine.

 

[Radost]

The XAS is a different machine.

Pretty much the same machine as far as scanning quality. Unless you are on a Apple silicon or really care about faster scanning there is no reason to upgrade.