Epson V750 vs Imacon 949

[joolsb]

Managed to get my example working on the other post but thought I'd copy it here..

Obviously there is a striking difference at full res when viewed on a computer monitor but would this be anywhere near as marked in an A3 print? Or even an A2? I'm not so sure. I think it makes much more sense to compare final output, to be honest. Over to you, Tim.

I have some very acceptable A2 prints made from Epson scans which stand up very well to nose-against-the-print inspection and I'm sure others have too. A drum scan needs less 'sharpening' (although I was surprised to find when working on one of Tim's scans that even drum scans can benefit from a little capture-sharpening) and less processing overall so there are definite advantages but the humble Epson doesn't really do that bad a job for the money.

[Joanna Carter]

An Epson printers's native screening (turning pixels to ink dots) resolution is 360ppi, or 720ppi if you're printing at the full highest res (2880dpi etc)

DJ, you seem to be confusing printing resolution in dpi with image resolution in ppi. It is commonly accepted that 240ppi is perfectly acceptable for printing to an inkjet printer, thus giving a 10x magnification if you scan at 2400ppi. It would be interesting to see if there is any perceivable difference in print quality if an image is printed, at the same scale, whether it is scanned at a ppi resolution, that is an exact mutiple of 360dpi, as compared with the "native" resolution of a scanner that uses 2400ppi. Personally, I don't believe the print dpi resolution should be an important factor when it comes to scaling the scanning resolution in ppi.

[AbsolutelyN]

In my comparison of an Epson v750 to a Howtek 7500 the Epson scan used the standard Epson 5x4 film holder as I did not have a better scanning holder. I believe Tim's comparison was wetmounted with the better mount.

[scovell001]

This is all really great stuff, thanks for all your comments.

I've got a proposal. If I hook out a tranny, can I send it to Tim (for your Howtek), AbsolutelyN (for your Howtek) & Joanna (for your V750 as your obviously well up to speed with getting the best from yours). I'll print the images, scan the prints & post them here for all to compare.

Does that sound like a plan? Our own collaborative large format scanner comparison.

[Charles Twist]

Hello Ian,
Sorry for butting in, but I would be interested in comparing the results to my V700, if at all possible. I don't know if it's possible, but could I get the tranny too at the end of the chain? I can either benchmark for myself or send the file on to you. Let me know.
Thanks and best regards,
Charles

[scovell001]

Sure Charles, lets do this.

Tim/Joanna, if your in can you give me a PM/post on this topic. As reiterated in my original posting, this isn't about being better than one another. This is a collaboration, to see where these machines rank, and more importantly can the difference be noticed in a print.

take care

I

[Joanna Carter]

I'm out of the country until tonight. I'll send something in the morning

[Thingy]

It sounds like a very good idea Ian. It would certainly be interesting to compare one of those legendary Scovell scans with what us mere mortals can achieve with our Epsons.

I look forward to the results.

[DJ]

DJ, you seem to be confusing printing resolution in dpi with image resolution in ppi.

Nope, I am very aware of the difference, if you re-read through my post you will see the two terms are used appropriately in their respective places.

PPI is the input resolution, DPI is the output resolution. The bit that turns one into the other is called "screening".

It is commonly accepted that 240ppi is perfectly acceptable for printing to an inkjet printer, thus giving a 10x magnification if you scan at 2400ppi.

It's a common misconception. The origin of the "240ppi" was based upon and old idea of dividing the DPI resolution by the number of inks (1440/6 = 240) because it was (mistakenly) assumed that the printer would lay down dots side by side, which is not the case. Unfortunately this gained momentum on internet forums and still perseveres today, there's no technical basis to the 240ppi magical figure other than plucking a number out of thin air.

The actual rasterizing or "screening" resolution of the Epson drivers is 360ppi (or 720ppi if working at the highest printing resolution), that's the native input resolution the screening algorithm works at. If you feed it an image at 240ppi, the driver will first upscale the image (using the cheapest algorithm possible, nearest neighbour) to 360ppi, and then screens it into ink dots.

If I can send an image to the printer at 360ppi I know I'm giving it as much data as it needs, if I send one at 240ppi I know it's not enough, and must accept that the driver will interpolate the rest. If the printer driver inventing data is acceptable to you, more power to your elbow, for me though, I'm going to send it as much as it can use.

It would be interesting to see if there is any perceivable difference in print quality if an image is printed, at the same scale, whether it is scanned at a ppi resolution, that is an exact mutiple of 360dpi, as compared with the "native" resolution of a scanner that uses 2400ppi. Personally, I don't believe the print dpi resolution should be an important factor when it comes to scaling the scanning resolution in ppi.

Linking the scanning resolution with the printing resolution in this way is an old throwback to the reprographic/printing industry, and the "this is how we've always done it" mentality that pervades it.

To make the optimal print, you should scan at the highest resolution possible ( taking into account native optical scanning resolution and diminishing returns ), then when ready to print, set the PPI of your master image to the native resolution of your printer's screening algorithm, and downscale the image to the size of print you want, then apply output sharpening at this final target resolution, then print.

[timparkin]

Sure Charles, lets do this.

Tim/Joanna, if your in can you give me a PM/post on this topic. As reiterated in my original posting, this isn't about being better than one another. This is a collaboration, to see where these machines rank, and more importantly can the difference be noticed in a print.

take care

I

Got the message and I look forward to getting the scan done!

Tim

[Dave Tolcher]

I am with Jools on this one and think that you need to close the loop and print the output, not just look at on screen where it will (should!) be a no brainer as to what captures the most and best information. I know that the output from my v700 is a bit soft and noisy in deep shadow but 25x20 prints are as good quality as anyone else's I have seen even at nose press level. At this size of print and for a well balanced slide I believe it is well past the law of diminishing returns to *need* anything better. At this stage I should say that they are on matt paper.... equation may well be different for a chromira print for eg.

What I struggle with is getting any quality out of a low key slide or one that I have underexposed by more than 1/2 stop particularly with velvia 50 where the shadows have blocked up. The V700 really struggles with anything more than 1.2-1.3 stops below 17% gray in my experience.

[timparkin]

I am with Jools on this one and think that you need to close the loop and print the output, not just look at on screen where it will (should!) be a no brainer as to what captures the most and best information. I know that the output from my v700 is a bit soft and noisy in deep shadow but 25x20 prints are as good quality as anyone else's I have seen even at nose press level. At this size of print and for a well balanced slide I believe it is well past the law of diminishing returns to *need* anything better. At this stage I should say that they are on matt paper.... equation may well be different for a chromira print for eg.

What I struggle with is getting any quality out of a low key slide or one that I have underexposed by more than 1/2 stop particularly with velvia 50 where the shadows have blocked up. The V700 really struggles with anything more than 1.2-1.3 stops below 17% gray in my experience.

I agree - I've seen Richard Child's prints from images produced on the Epson and they are fantastic and you don't "need" anything better (but then again, we don't need to take pictures or need to use large format cameras so that may be a wrong approach to assess this by).

My reason for getting the Howtek was for the deep shadows more than the resolution. The resolution is a 'bonus' in that if I ever did do a show and someone asked for a 24x20 or 30x40, my original, post processed scan would be ready to go.

However, we're talking about whether there would be 'any visible improvement' in using a drum scan (virtual or not) and at what print size it starts to be 'desired' or 'required'. I'm really looking forward to the results.

Tim

p.s. I think a chromira print would be need the higher resolution and deep blacks even less. A matt inkjet print is about as sharp as you can get and exposes as much dynamic range too..

[Joanna Carter]

Nope, I am very aware of the difference, if you re-read through my post you will see the two terms are used appropriately in their respective places.

I would expect nothing less from you

The actual rasterizing or "screening" resolution of the Epson drivers is 360ppi (or 720ppi if working at the highest printing resolution), that's the native input resolution the screening algorithm works at. If you feed it an image at 240ppi, the driver will first upscale the image (using the cheapest algorithm possible, nearest neighbour) to 360ppi, and then screens it into ink dots.

If I can send an image to the printer at 360ppi I know I'm giving it as much data as it needs, if I send one at 240ppi I know it's not enough, and must accept that the driver will interpolate the rest. If the printer driver inventing data is acceptable to you, more power to your elbow, for me though, I'm going to send it as much as it can use.

...

To make the optimal print, you should scan at the highest resolution possible ( taking into account native optical scanning resolution and diminishing returns ), then when ready to print, set the PPI of your master image to the native resolution of your printer's screening algorithm, and downscale the image to the size of print you want, then apply output sharpening at this final target resolution, then print.

Being a somewhat pedantic person I would question the logic behind this, as the pixels in the image are square but the inkjet dots are (approximately) round(ish). No matter what the "native resolution" of a printer, it still has to make a best guess as to where to place the relevant dots to make up the square pixels. There is also the question of how the printer driver would render a pixel of a colour that requires a mixture of the five/seven/nine colours that make up the cartridge set.

I actually wonder if you are simply replacing one magical number with another

[Charles Twist]

Do you mind, Tim? I have just coughed tea all over the monitor. Don't NEED to take pictures; don't NEED to use large format?!!? What planet are you from? Are you normal?
As for the rest, I agree heartily that shadows are an issue. As I said earlier, it's the blasted green noise. Now, does anyone know where I can get a drum scan done for a tenner? I really think this test should be done with my huge pile of dark trannies (film noir, some call it). In fact, all joking aside, it might be quite interesting to carry this test out with a dark tranny.
Regards,
Charles

[Dave Tolcher]

I believe that David Meunch on the IOW will do drum scans for around the £15.00 mark (last time I checked). He does scan and print for the 'Wardian' clan I think (Anna Booth, David Ward and a few others) - all the work exhibited at the Oxo gallery in their recent exhibition was scanned & printed by David. He doesnt have a website but I have a phone number if you want me to pm it across (I'll post here if allowed by the forum rules).