Your Workflow for Large M8 Prints?

[buckhorn_cortez]

Advertisement (gone after registration)

I've been making 20x30 inch prints from my M8, and over the past 3 months have tested the following resolution interpolation software: Genuine Fractals V, PhotoZoom Pro 2, Imagener, Humansoft Xpline, and Qimage.

 

The only workflow I've found that produces an artifact free, high resolution image is: processing in Capture One 4 with final tweaks in PS in 16 bits. Converting the 16-bit file to 8 bits, processing the 8-bit file with Noiseware; and then printing the file through Qimage with an Epson 9800.

 

Anyone have another workflow that they've tried and would like to suggest?

[canlogic]

I have been using both Aperture and mostly Lightroom. I was round tripping to CS3 and using Photokit sharpener for both input and output sharpening. With the new version of LR and the fact that most if not all of Photokit sharpener is now incorporated into it I do almost everything directly in it.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Shootist]

I have been using both Aperture and mostly Lightroom. I was round tripping to CS3 and using Photokit sharpener for both input and output sharpening. With the new version of LR and the fact that most if not all of Photokit sharpener is now incorporated into it I do almost everything directly in it.

 

Could you explain why you think the Photokit sharpener is now included in LR2. The only sharpening I see in LR 1 or 2 is in the RAW converter and whether you are using the stand alone ACR or LR 1 or 2 they are the same.

[canlogic]

Jeff Shewe one of the owners of Pixel Genius worked very closely with Adobe on the input and output sharpening in LR 2. In a couple of posts from him I believe he has even stated that sharpening is better in LR2. He posts fairly often on Luminous Landscape.

[biglouis]

I've been making 20x30 inch prints from my M8, and over the past 3 months have tested the following resolution interpolation software: Genuine Fractals V, PhotoZoom Pro 2, Imagener, Humansoft Xpline, and Qimage.

 

The only workflow I've found that produces an artifact free, high resolution image is: processing in Capture One 4 with final tweaks in PS in 16 bits. Converting the 16-bit file to 8 bits, processing the 8-bit file with Noiseware; and then printing the file through Qimage with an Epson 9800.

 

Anyone have another workflow that they've tried and would like to suggest?

 

I've only printed up to A3 on a R2400 but I've managed frighteningly sharp images (I mean images where the detail is so fine I'm surprised the camera was even able to capture it, let alone the printer print it).

 

My workflow is as follows - RAW conversion in Lightroom with adjustment to black point and clarity --> CS3 where I adjust curves, convert to LAB colour and boost colours via soft light if needed, adjust shadow and highlight if needed, adjust images size and resolution if needed, adjust and fade sharpening --> convert to RGB --> Print.

 

I envy you your 9800 although I doubt I'd have much use for larger than A3 prints. With the software you have in your workflow I'd expect excellent results, although as with all things, GIGO rules (garbage in, garbage out).

 

LouisB

[buckhorn_cortez]

"With the software you have in your workflow I'd expect excellent results, although as with all things, GIGO rules (garbage in, garbage out)."

 

The problem in making a large print seems to be to find out how best to interpolate the image to the final print size. In some cases, the interpolation software gets confused (Genuine Fractals being the prime example - but not the only one) with small details (crossing branches, for example) - and may turn them into 3D objects by filling the area between the branches.

 

In other cases, the interpolation software will give artifacts on totally flat colors. As an example, there will be a group of pixels as a block of color that is out-of-place within the flat color field.

 

Using either straight bicubic, or stepped bicubic is the least satisfactory as the edge sharpness and details are lost.

 

I've found that the Noiseware step on the file prior to printing with very slight sharpening helps in several ways. The first is that even with Qimage as the file formatting / interpolation software, if you look very closely at flat color areas (blue sky for example) - you can see a very minute pixel-based uneveness. I've found that by using noise reduction on the file prior to sending through Qimage, that this slight uneveness goes away. This also results in more detail in the final print as small lines (like telephone or power lines) are more solidly rendered giving the impression of higher resolution.

 

In testing both FocalBlade and Noiseware, I've found that for the M8 images, Noiseware is less destructive (doesn't soften the image) like FocalBlade and can provide a tiny bit of sharpening prior to final sharpening in Qimage. For reference, I have my M8 set with sharpening "off."

[charlesphoto99]

Advertisement (gone after registration)

I do 10-20% increments in CS3, bicubic standard. I tested against Alien Skin Blowup and GF. It's the BIG jumps that get you in trouble. Do a lot of little ones and it will fill in each time only what is needed. Give it a try.

 

ps; this was suggested to me by a well respected drum scanning house I use. They said that's how they do it for resizing both up and down.

[Guest DuquesneG]

For reference, I have my M8 set with sharpening "off."

 

After all your learned discussion I am surprised at that comment. AFAIK the M8 doesn't apply sharpening to DNGs, so then, you are shooting JPEGs FWIW I don't use any up-rezing, sharpening, or noise reduction (unless I'm shooting above 640, which I try never to do). I have them all turned off in C4 and I just print straight from the output. I haven't made a ton of 20x30s but enough to feel that's all they need. Of course nobody I know carries a loupe in their pocket and examines my prints from a 1/4 inch away. Normal viewers stand a few feet away from a print that size. This is a case where uploaded files aren't much help/proof, because you just know people are going to pixel peep.

[buckhorn_cortez]

AFAIK the M8 doesn't apply sharpening to DNGs, so then, you are shooting JPEGs

 

I understand the the DNG is not sharpened, I should have clarified WHY I have sharpening turned off - my mistake. I like to compare the DNG image to the unsharpened JPEG - it gives me a little more information about the individual image beyond what I see in the DNG on the monitor - it's a point of reference only. I find the JPEGS from the M8 unusable for anything else, and if I need a JPEG, I make it from the final, corrected image.

 

FWIW I don't use any up-rezing, sharpening, or noise reduction (unless I'm shooting above 640, which I try never to do). I have them all turned off in C4 and I just print straight from the output. I haven't made a ton of 20x30s but enough to feel that's all they need.

 

I'm not sure how you can do that and get a quality print. At 20 x 30 the image is about 130-140 dpi (I'd have to do the math to get it exact) - and you're letting the printer driver do the resampling - which they are not all that good at. Are you using a RIP and bypassing the printer driver? Qimage acts somewhat like a RIP on output as it resamples and reformats the image, then sends the printer the optimum pixel matrix to match the printer's native resolution - for my Epson 9800, that would be 720. This eliminates the printer driver resampling.

 

Of course nobody I know carries a loupe in their pocket and examines my prints from a 1/4 inch away. Normal viewers stand a few feet away from a print that size. This is a case where uploaded files aren't much help/proof, because you just know people are going to pixel peep.

 

I'm not sure what constitutes a "normal viewer." Is there an ANSI or ISO standard that defines this? (for joke). I like to look at photographs, other types of prints (lithographs, intaglio,etc.), paintings, sculptures, and other art closely to see what the surface looks like, and what kind of details are being shown.

 

I'm especially interested with inkjet prints with the ink interaction with the paper surface. So, I get up close to see how the ink lays on the paper. In doing that, I become aware of how the image is presented from the surface to viewing distance.

 

If I'm happy with one of my prints looking at it from 6-10 inches from the surface, when I stand back away from the print to normal viewing distance it seems to look even better.

 

I do 10-20% increments in CS3, bicubic standard. I tested against Alien Skin Blowup and GF. It's the BIG jumps that get you in trouble. Do a lot of little ones and it will fill in each time only what is needed. Give it a try.

 

As I've stated previously, I've tried single step and incremented bicubic interpolation. I have a PS macro that does 10% increments. When you get past about 13x19 in size, that method is just too soft and pixelated.

[Guest DuquesneG]

I understand the the DNG is not sharpened, I should have clarified WHY I have sharpening turned off - my mistake. I like to compare the DNG image to the unsharpened JPEG - it gives me a little more information about the individual image beyond what I see in the DNG on the monitor - it's a point of reference only. I find the JPEGS from the M8 unusable for anything else, and if I need a JPEG, I make it from the final, corrected image.

 

Gotcha.

 

 

 

I'm not sure how you can do that and get a quality print. At 20 x 30 the image is about 130-140 dpi (I'd have to do the math to get it exact) - and you're letting the printer driver do the resampling - which they are not all that good at. Are you using a RIP and bypassing the printer driver? Qimage acts somewhat like a RIP on output as it resamples and reformats the image, then sends the printer the optimum pixel matrix to match the printer's native resolution - for my Epson 9800, that would be 720. This eliminates the printer driver resampling.

 

I export anything that large to a professional printing company, so I can't answer your question. I know they aren't up-rezing or sharpening my files because they told me so. They found out after doing a bunch of prints from M8 files that it hurts more than helps.

 

 

 

I'm not sure what constitutes a "normal viewer." Is there an ANSI or ISO standard that defines this? (for joke). I like to look at photographs, other types of prints (lithographs, intaglio,etc.), paintings, sculptures, and other art closely to see what the surface looks like, and what kind of details are being shown.

 

I'm especially interested with inkjet prints with the ink interaction with the paper surface. So, I get up close to see how the ink lays on the paper. In doing that, I become aware of how the image is presented from the surface to viewing distance.

 

If I'm happy with one of my prints looking at it from 6-10 inches from the surface, when I stand back away from the print to normal viewing distance it seems to look even better.

 

That makes sense. My 20x30 prints from M8 files look better at 6 inches, without sharpening or up-rezing, than they do with. When I first got the M8 I fussed around with plugins too, sticking to the theory that 10 megapixel files would need up-rezing and sharpening to print at 20x30, and I was disappointed with the results. Once the printer place had done a number of 20x30s from M8 files, it was them who told me to quit all the post processing tweaks. Just following orders here

[Printmaker]

I use Alien Skin Blowup for everything 20x30 and up from M8 DNG files which were converted to 16 bit TIFF in Capture One. No sharpening until everything else is ready, then Photoshop USM (30/20/1 or 120/1/1). I get artifacts if I sharpen too early in the file prep. These artifacts are easy to see from contrasty Zeiss lenses and less obvious from 90s vintage Leica lenses.

 

Tom

[buckhorn_cortez]

I export anything that large to a professional printing company, so I can't answer your question. I know they aren't up-rezing or sharpening my files because they told me so. They found out after doing a bunch of prints from M8 files that it hurts more than helps.

 

If it's a professional printing company, I'm sure they're using a RIP (if it's an inkjet print) which will interpolate the image to the correct amount of data (file size) needed to support the image size. If you're having prints done on a LightJet or Durst Lambda, the machines have their own built-in interpolation system.

 

If they're using a RIP for inkjet, the RIP will provide sharpening as part of the process. That way you sharpen correctly for the size of the image on output. Likewise, with the LightJet or Lamda, sharpening is part of the process as the printer formats the print for the final output size.

 

That's why they told you to not sharpen or post process your image - either their software RIP or printer is doing the sharpening. Qimage has sharpening built-in, and like a RIP will sharpen on output. The tiny amount of presharpening I'm applying for this series of prints seems to enhance the final print, as I've evaluated prints side-by-side with no sharpening and a small amount applied as part of the noise reduction.

 

The ones with the small amount of sharpening applied pre-printing seem to hold a tiny bit more detail in things like lettering (signs), textures, etc.

[buckhorn_cortez]

Buckhorn - I use Qimage too (Studio Edition), and really like it too. What interpolation settings do you use? After experimenting, I found that the following gives me maximum detail with minimum artifacts:

 

• Pyramid interpolation on default settings

• Final print sharpening: 4 (one notch down from the default), smart sharpening on.

 

(I'll be trying out your noise reduction step, as that may change my interpolation method. My current workflow is to avoid noise reduction on lower-ISO images as much as possible to retain maximum detail - but from what you say, very fine detail may actually be better off lost when printing large [and won't matter if printing small], to create a more coherent fine structure to the image.)

 

For the large prints I use the pyramid interpolation with print sharpening at 5 & smart sharpening "on." As for the noise reduction, I've made prints with no noise reduction and from noise reduced files. I find the prints with noise reduction are better when using Noiseware. Noiseware seems much less destructive to the image than Neat Image. Noiseware has a lot more controls that can be adjusted so you can get the image noise out with no degradation (softening) of details. I used Neat Image for about 5 years, and switched to Noiseware earlier this year when I started using the M8. Neat Image was okay for scanned film, but I found it too agressive for digital camera images.

 

Noiseware allows you to address luminance noise separately from color noise, and you can tweak the noise settings by individual color noise or by tonal range (white, greys, black). There is a "sharpen" function (called "Detail Enhancement") in Noiseware that is complementary to the noise reduction; and adding a small bit of sharpening as part of noise processing seems to give more detail in the final print. I usually set the Detail Enhancement at 3.

 

Like most all of this kind of sofware there is a "pre" and "post" processing split screen so that you can see the effect of the software on the image. I usually put the monitoring window over a finely detailed area of the image to make sure there is no detail destruction as part of the processing. The software provides an automatic analysis of the image, and "learns" from each image as you tweak the controls. After about 4 images, it has the settings learned, and I've found I no longer have to manually work the image, I only move the monitoring box around the image and check areas to ensure it's the way I want it to look after processing.

[haroldp]

My typical workflow is:

 

DNG processing in Photoshop / ACR with final tweaks in PS in 16 bits. Converting the 16-bit file to 8 bits (JPG highest quality), and then printing the file through Qimage (hybrid interpolation, 600 dpi, final sharpening at 4) with an HP Z3100.

 

I rarely shoot over ISO 320 on the M8, ( I never used film faster than 400 ), so I have not found post noise reduction to be useful to me. When high ISO's are needed my D3 comes out.

 

I will of course vary the above if my situation is not 'typical' for me.

 

I use Photoshop / ACR because I also use other raw formats as well, and get confused with to many tools.

 

I hope this is helpful.

 

Regards .... Harold