Plustek???

[Ivo Stunga]

Note to mods and @snusmumriken - I was calling myself illiterate (in scanners) - which I am, therefore I have to rely on a side-by-side comparison instead.

Have done it with 2 completely different prints and have to agree with various reviews and tests out there: Plustek doesn't do 7200. End of story.

I'll repeat myself - I see no improvements having a A2 print from PlusTek 7600i @3600 on the table next to 7200 - virtually identical and no extra details to be had at "7200" - I mean - we're talking "DOUBLING" the resolving power, so I should at least see some improvement...
Does 3600 scan maxes out Delta 100 and Aviphot 200?

Hmm, to be extra sure I could even repeat this test for the third time with more resolving film - like with Aviphot 80

 

[koraks]

I was calling myself illiterate (in scanners)

The formulation of your post didn't make that explicit, so I think that was why it was binned. Don't worry about it, but keep in mind that when using terms like 'illiterate' this way, there's a chance people will misunderstand even if you mean well.

 

[snusmumriken]

Erwin Puts had some pics of what grain looks like at magnification and as Helge noted it would take a lot of pixels to effectively duplicate the larger magnifications:

basic facts about film

Projection magnification may be better than present day scanners at duplicating silver grains.

The ship Amateur Photographer used is still there, it was more readable in those days.

Yes, I already understood most of what Puts wrote there (and he managed not to mention Leica once, although the red and black text is a bit suggestive!)

What I haven’t completely grasped is how a scanner operating at low resolution creates a noise pattern that is irregular like graininess, and presumably founded on the same grain clumping, but coarser and less distinct than what you see optically.

@Ivo Stunga This incivility thing has passed me by, and I don’t know how I am implicated. I hope I haven’t inadvertently caused offence somewhere?

 

[koraks]

I hope I haven’t inadvertently caused offence somewhere?

No, you didn't. Nobody did. It was all just a small misunderstanding. Don't worry about it!

 

[Alan Johnson]

What I haven’t completely grasped is how a scanner operating at low resolution creates a noise pattern that is irregular like graininess, and presumably founded on the same grain clumping, but coarser and less distinct than what you see optically.

Join the club, this article is from 2000:

Grain aliasing

 

[Ivo Stunga]

The formulation of your post didn't make that explicit, so I think that was why it was binned

Yeah, it was in the context of my previous post, therefore easy to misunderstand.

I hope I haven’t inadvertently caused offence somewhere?

You haven't, all is good. It's just that my tests (of physical prints side-by-side) show what they show - don't know why Plustek claims otherwise.

 

[brbo]

It's just that my tests (of physical prints side-by-side) show what they show - don't know why Plustek claims otherwise.

They claim true optical resolution of 7200dpi. By that they obviously mean a sensor capable of outputting at least 7200px per inch and a stepper motor capable of 7200 steps per inch. I don't see them promising effective resolution of 7200dpi. Claiming optical resolution of 7200dpi is disingenuous, but maybe not straight out lying. Many manufacturers nowadays do that.

If you bought your Plustek and really think that they should deliver effective 7200dpi for that money, return the scanner. Maybe if all customers did that, this practice would change. That said, you should know that no scanner that can fit your room can deliver effective 7200dpi. You could've saved a bit of money and get a bit higher resolutions if you bought an older Minolta/Canon/Microtek scanner, but if you wanted a scanner with warranty and modern connectivity... Plustek is not the worst buy.

 

[snusmumriken]

Join the club, this article is from 2000:

Grain aliasing

Despite being inconclusive, that's a really helpful article. Thanks!

 

[MattKing]

Despite being inconclusive, that's a really helpful article. Thanks!

In some ways that describes a lot of Photrio threads !

 

[snusmumriken]

In some ways that describes a lot of Photrio threads !

 

[Alan Johnson]

In some ways that describes a lot of Photrio threads !

 

[Brad Deputy]

Here's a comparison of 7200 vs 3600 DPI on a Plustek 8200. Left to right respectively.
Adox HR-50, HR-DEV, Olympus 35 RC

Not the sharpest camera around, but I can see some details missing. For example, in the "4" in 142 I can almost make out the space inside.
I used the "Less Auto Sharpening" setting under Unsharp Mask.
Full text reads "142 ST SW, DEAD END -->".

For my work, it's not quite enough gain in detail to scan everything at 7200 DPI (with 4x file sizes). But I feel more detail is to be had. The "4200" or so DPI guess made by others bears some consideration.

 

[Anon Ymous]

Here's a comparison of 7200 vs 3600 DPI on a Plustek 8200. Left to right respectively.
Adox HR-50, HR-DEV, Olympus 35 RC
View attachment 350345

Not the sharpest camera around, but I can see some details missing. For example, in the "4" in 142 I can almost make out the space inside.
I used the "Less Auto Sharpening" setting under Unsharp Mask.
Full text reads "142 ST SW, DEAD END -->".

For my work, it's not quite enough gain in detail to scan everything at 7200 DPI (with 4x file sizes). But I feel more detail is to be had. The "4200" or so DPI guess made by others bears some consideration.

Somehow, all I see is huge pixels. I can't read any text in any of the images. Am I the only one?

 

[snusmumriken]

Somehow, all I see is huge pixels. I can't read any text in any of the images. Am I the only one?

I can see bigger pixels on the right than on the left, and by an act of faith I can believe that the blob group at the top left of each reads 142. Inclined to agree with @Brad Deputy ’s conclusions.

 

[Anon Ymous]

... and by an act of faith I can believe that the blob group at the top left of each reads 142...

Ah, yes, noticed that now at the higher resolution image.

 

[koraks]

If you zoom back out, it becomes (sort of) visible:

I'm not convinced there's actual information gain going from the 3600dpi to the 7200dpi scan. There are more pixels, but it seems to be a combination of interpolation a lower s/n ratio.

For example, in the "4" in 142 I can almost make out the space inside.

Emphasis on 'almost' - you can 'almost' make it out if you already know it's there.

Whatever additional information is actually captured, is so little that its benefit is IMO marginal at best.

 

[Ivo Stunga]

Yeah, an "improvement", costing you an afternoon to achieve it

 

[Alan Edward Klein]

What would be interesting is if you had a third sample that took the 3600 original, and uprezed it to 7200 by an editing program.

 

[koraks]

What would be interesting is if you had a third sample that took the 3600 original, and uprezed it to 7200 by an editing program.

The result will depend heavily on what kind of upsampling tool is used. Here's a fairly random example:

Of course, as with all AI, an old principle from the IT world applies - one that was reiterated and emphasized in a presentation of a professor in AI I attended before the weekend: GIGO. Garbage In = Garbage Out.