Questions about Qimage

How many of you are using Qimage, and why?

Does Qimage actually produce any better print quality that CS3?

I know a lot of people use Qimage, but I have never tried it, and would like to see some opinions before looking more seriously at it.

I am pleased with my prints currently, but always willing to look at anything that might improve them.

In case in makes a difference with using Qimage, I shoot a D2x, use Capture 1 for RAW conversion, CS3, PK Sharpener, Jos. Holmes Dmax3 colorspace, and a Chromix custom profile for Moab Entrada Natural, and print using an R2400.

Thanks for any input and/or opinions.

Bill, your best bet is to download a trial version and try it yourself. I have never understood the somewhat "religious" following this program has, but those that have adopted it into their workflow seem to love it. (and aren't shy about stating such! ) My own trials of it awhile ago (comparing prints, not screen shots) showed no better for enlargements than using Bicubic Smoother and some sharpening afterwards. Every comparison I have seen wasn't really telling the whole story. But it does offer you benefits if you are trying to layout multiple prints on a page. But hey, if it works for someone - go for it.

Timely question, Bill. I'll be interested to see some additional responses. I downloaded Qimage a few months ago, but I was so busy during most of the month after I installed it that I never really had a chance to print with it before the trial ran out (I subsequently uninstalled it). On initial inspection, I found it about as intuitive to use as Photoshop was when I first started messing with that application five or six years ago. I've heard a number of people claim that it provides enlarged prints that are significantly better than what can be created using PS alone. I suppose the question I have for those folks is how much larger they feel then can print with acceptable results using Qimage rather than PS. If there's a material enlargement benefit to using Qimage, I'd be willing to take another look, but I'd like some sense of the scale we're talking about.

I have written about Qimage here from time to time, and there are some others here who use it also. If you haven't yet done so, search the archives for "qimage" and you'll find some relevant discussions. I won't try to repeat all of the past details of why I use it. But to summarize, there are a few key points:

• - Interpolation: Qimage has multiple interpolation algorithms, several of which produce (to my eye) better looking results when scaling up for printing large, compared to Bicubic. Some of Q's algorithms can be tuned for performance vs. quality, others not; but the choice is there. You would have to evaluate some of your own images at final resolution to determine whether you feel the results are better. For me, I prefer Q's Vector, Pyramid or Hybrid interpolation over Bicubic.
  - Sharpening: Most people don't use a multi stage sharpening workflow, applying a level of post-interpolation sharpening a la PK Sharpener. For these people, Q will offer an immediate benefit because it provides the option to combine some degrees of pre-interpolation sharpening, interpolation to the appropriate resolution for the printer, and then post-interpolation "smart" sharpening that varies based on how much the image had to be sized up. If a person already uses PKS or some other multi stage sharpening workflow, Q won't necessarily be able to sharpen better, but it may do so with less manual work.
  - Workflow: Q is like a "swiss army knife" of printing, and has all kinds of functionality. From automatic scaling up to printer resolution, to application of adjustments using non-destructive associative filters, to handling multiple images per page, to flexible setup/save/recall of common print jobs, etc. -- the more you print and the more kinds of printing you do, likely the more time Q will save in your workflow. Conversely if you only print small jobs on occasion, single image per page, with limited variations, then you won't be tapping into a lot of Q's functionality. At this point you mainly would decide its benefits based on the above points plus the fact that it doesn't force you to save large interpolated files.

Kerry -- Q definitely has a user interface that is an "acquired taste". I don't know that I'd characterize Q's benefit as "how much larger" one can print with acceptable results. To me that implies size is the driving thing. But it isn't for me, and I don't evaluate Q on that basis. So I can't directly respond to your interest in that sort of measurement. I can say Q makes an observable qualitative difference in the sizes of printing I commonly do -- single sheets up to 13x19, and large pano's on roll stock up to several feet long. It also simplifies some printing workflow for me, reducing the amount of laborious interpolation, output sharpening, and giant file management that I have to deal with.

For somebody already using a workflow as tuned as Bill's is or yours might be, the bang for buck may be less... "compelling". You'd have to want to invest the effort to get an incremental improvement. Without knowing a lot more specifics about the level of quality any of you are getting now, and how acceptable that is for your purposes, it's basically impossible to say what scale of improvement you might be looking at. It certainly isn't going to be a night & day difference... but it might be something on the order of 5 - 10% in certain qualitative areas like detail & perceived sharpness. Manual effort reduction might be more significant than that, once the app learning curve is dealt with. And then if you need any of the "swiss army knife" functionality, that starts to add more value beyond the level of just making single prints...

Thanks Royce for the superb comment. One question, and this goes back to the question I was asking you & Eric last week, RE: output resolution for the R2400 being 720. I currently print at 360 resolution and decided to continue that because of the HUGE file size difference between 360 and 720. Am I understanding you correctly that with Qimage, I can still save at 360, and Q will upsize to 720 before it sends it to the printer?

If so, that alone would probably be worth the price, which is very reasonable anyway.

Yes, technically speaking you could use your current workflow with a manual interpolation up to 360 dpi and PKS for post-interpolation sharpening in Photoshop. Q would then (normally) step this up to 720 dpi and apply its own post-interpolation sharpening before sending the data on to the printer driver. This may or may not be desirable, because you're using a compound interpolation & print stage sharpening workflow where the 2 approaches may not interact very well with each other.

Of course you can instruct Q not to interpolate (or interpolate to a lower resolution, the next smallest of which would be 360 dpi where you are now). In this case it will just send your file as-is -- no additional interpolation, and no Q "smart sharpening". However at this point unless the "swiss army knife" printing functions of Qimage are useful to you, one would have to question why you are bothering with it.

With Q in the loop for the output stage, personally I would recommend doing without any interpolation or post-interp sharpening in Photoshop. Let Q take the file all the way from regular camera resolution up to 720 dpi for the printer. The would mean leaving PKS out of the equation for output sharpening, and letting Q manage that as well. Evaluate the results of doing this vs. your current workflow based on PKS and 360 dpi. If you find the results work at least as well with Q by itself, then you'd be saving yourself a manual step with PKS and the big(ish) 360 dpi file.

OK. I think I understand all you have said! Just to make sure though, please allow me to repeat back to you.

Use PKS for capture and intermediate sharpening, but not for output sharpening, correct?

Instead, allow Q to resize and apply ouput sharpening for each print size, correct?

One part I'm fuzzy on concerns your last sentence ("... saving yourself the big(ish) 360 dpi file"). My default print resolution in CS3 is 360, screen resolution is 72. Are you saying to change my default print resolution to a smaller size, and allow Q to up-rez all the way to 720? If so, what default printer resolution do you suggest, or does it matter?

My interest is definitely piqued enough so that I will download the Trial Version and check it out.

Will Q allow you to soft proof, or do you still use CS3 for that?

Thanks again, you've been a huge help!

Qimage can soft proof...

Bill Chambers wrote:OK. I think I understand all you have said! Just to make sure though, please allow me to repeat back to you.

Use PKS for capture and intermediate sharpening, but not for output sharpening, correct?

Instead, allow Q to resize and apply ouput sharpening for each print size, correct?

Yes, you got that right.

One part I'm fuzzy on concerns your last sentence ("... saving yourself the big(ish) 360 dpi file"). My default print resolution in CS3 is 360, screen resolution is 72. Are you saying to change my default print resolution to a smaller size, and allow Q to up-rez all the way to 720? If so, what default printer resolution do you suggest, or does it matter?

Printing is a bit different in CS3 than CS2, and I haven't installed my copy of CS3 yet. So my terminology may be a bit off based on what you're seeing in CS3 dialog boxes.

What I mean is, do not interpolate the file in any way in Photoshop. It has however many pixels it has, coming out of the camera. Just leave it like that. Qimage will interpolate the file from whatever resolution it is, up to 720 dpi for the printer. Or a lesser value, if you choose 360 dpi for example. But I'd say try it at 720 first since Qimage does not save the massive interpolated file to disk. And as we discussed before if excessive consumption of disk space is eliminated as a downside, going to 720 dpi may show some noticeable improvement in detail for you.

Will Q allow you to soft proof, or do you still use CS3 for that?

Lou answered this -- Qimage can soft proof after a fashion. But to elaborate, I prefer Photoshop for that purpose. Q's soft proofing is more rudimentary. It doesn't have a gamut warning function like PS does, for example. And if you soft proofed in Q and found an issue in the image, you'd just have to go back to PS anyway to correct it. So no real benefit in using Q to soft proof IMO.

Thanks again, you've been a huge help!

Either that, or I just helped you decide to waste a bunch of time!

Very helpful Royce, as always. I really appreciate your input.

howlandr wrote:Kerry -- Q definitely has a user interface that is an "acquired taste". I don't know that I'd characterize Q's benefit as "how much larger" one can print with acceptable results. To me that implies size is the driving thing. But it isn't for me, and I don't evaluate Q on that basis. So I can't directly respond to your interest in that sort of measurement. I can say Q makes an observable qualitative difference in the sizes of printing I commonly do -- single sheets up to 13x19, and large pano's on roll stock up to several feet long. It also simplifies some printing workflow for me, reducing the amount of laborious interpolation, output sharpening, and giant file management that I have to deal with.

For somebody already using a workflow as tuned as Bill's is or yours might be, the bang for buck may be less... "compelling". You'd have to want to invest the effort to get an incremental improvement. Without knowing a lot more specifics about the level of quality any of you are getting now, and how acceptable that is for your purposes, it's basically impossible to say what scale of improvement you might be looking at. It certainly isn't going to be a night & day difference... but it might be something on the order of 5 - 10% in certain qualitative areas like detail & perceived sharpness. Manual effort reduction might be more significant than that, once the app learning curve is dealt with. And then if you need any of the "swiss army knife" functionality, that starts to add more value beyond the level of just making single prints...

Thanks for the reply, Royce. Allow me to add a few specifics to see if that allows you to add some granularity to your recommendation. I currently shoot with a Nikon D200; landscapes make up the overwhelming majority of my subject matter with flora close-ups virtually all of the remainder. I've been completely happy printing anything uncropped--and I tend to do almost all of my cropping in the camera--from the D200 at 12x18 on my Epson 2200. This is essentially as large as I can print on sheets without going borderless. Frankly, I'd like to see how much larger I can print D200 images to my satisfaction, and one of these days I'll try upsizing to something like 16x24 and printing a cropped section on the 2200 for my own edification.

The real issue I'm facing is printing older images--ones captured with a D100 and its considerably less robust sensor. I've printed D100 images at 10x15 with reasonable satisfaction, but the end product really starts to break down--IMO--at 12x18, based on my experience. I guess I'm wondering if there's reason to believe that Qimage would produce an improved 12x18 print.

FWIW, in the past few months I've adopted a three-stage sharpening procedure that more or less mimics the PK approach (manually implemented, with heavy use of actions) and I interpolate using Bicubic Smoother in PS2.

If you--or anyone else--has any thoughts on this matter, I'd love to hear them.

Thanks.

Okay, that provides some parameters. I've used Qimage to print images from my old 6MP Canon 10D's on 13x19 stock and have found the quality acceptable. I don't know how my standards may compare to yours. I adopted Q back in the day when I wasn't following a full blown PKS style of sharpening workflow with manual interpolation up to printer res, etc. So the quality jump was pretty good for me at the time.

Given that you're already applying a good sharpening workflow and so on, Q's benefit to you would be less of a jump, and likely would be mostly visible where it does occur in images with lots of fine detail printed at 13x19 and larger. I'd expect this to be more a concern with your macro flora work than with your landscapes. I'm checking to see if I can quickly find a sample image that I might use for illustration with some 100% crops.

But for my purposes, I would say good files from a 6MP type camera should be quite doable at 13x19 media.

howlandr wrote:Okay, that provides some parameters. I've used Qimage to print images from my old 6MP Canon 10D's on 13x19 stock and have found the quality acceptable. I don't know how my standards may compare to yours.

Yeah, that really is the great unknown, unmeasurable factor in all of this, isn't it?

Given that you're already applying a good sharpening workflow and so on, Q's benefit to you would be less of a jump, and likely would be mostly visible where it does occur in images with lots of fine detail printed at 13x19 and larger. I'd expect this to be more a concern with your macro flora work than with your landscapes. I'm checking to see if I can quickly find a sample image that I might use for illustration with some 100% crops.

But for my purposes, I would say good files from a 6MP type camera should be quite doable at 13x19 media.

Which takes us back to comparative standards of acceptability.

Last month, I sold a 12x18 print of this image, which was a D100 capture. I thought that the quality of the print was borderline acceptable--as long as you maintained a "normal" viewing distance, it was fine. If you started to get your nose within a couple of inches of the print, it started to break down. My D200 images don't break down at all at 12x18, no matter how close you get to it. Now, in the case of the linked image, the customer was thrilled with it; I actually ran a print of it for her approval (we'd have gone 10x15 if she was dissatisfied), which I guess is all that matters. But I was left to wonder if it wasn't possible to do better.

Yes, quality is partly subjective and context dependent. E.g. will the customer notice it & pay extra for better quality, or reject an image with lesser quality? If not, why bother? Which is unfortunately why I often sound like a broken record in recommending that people do a serious eval to find out if they like or need a certain technique.

Looking at some things here (none of which I'm going to post for eval ) I'd say Q's Hybrid interpolation method does a better job of preserving solid looking edges in fine detail without artifacts, compared to PS Bicubic Smoother with some post-interp sharpening (using high pass which minimizes halos). The edges with Bicubic seem a little fuzzed out as if by dialing up the anti-aliasing, whereas the Q Hybrid edges are well defined. Q also seems to kick up unuseful "pseudo detail" in the shadow tones to a lesser extent without having to do anything special with it. Possibly something going on there like the Smart Sharpening filter in PS, which can roll off sharpening in the shadows; I could use the SS filter for post-interp sharpening instead of high pass but then might have to fight halos a little more.

I'm evaluating these differences by loading versions of images such as lizard scales & bird feathers as layers in PS, zooming to 50%, and then flicking between the layers to have a look at the deltas between them. The images are all uprezzed to 720 dpi for printing at a nominal 13x19 size. If I sized up only to 360 dpi in my PS version, Q would have a bit more advantage but I wanted to compare apples to apples.

In all fairness I must say that Bicubic Smoother, multi-stage sharpening and everything else I have learned in the past 2 years produces much closer results to Q now than was the case when I first started using it for my printing. But in the labor saving department, I still find it's less effort on my part to let Q handle the output side of the deal once I've prepped the image for creative purposes, at camera resolution. Handling all output steps with Q becomes very easy in comparison, with little manual effort beyond selecting the media's physical size.

I'm going to have to bite the bullet and do some experimenting followed by some good old-fashioned eyeballing to determine whether or not I see a substantial difference in output.

Thanks for all the help!

Thanks for the input, Royce. It has been some time since I've looked at QImage's interpolation methods. The Hybrid method is only available in the "pro" version, correct?

Yes, Hybrid is only available with the Pro edition. Awhile back, Mike Chaney also introduced a higher edition called Studio, which I use. It contains a variation of Hybrid called Hybrid SE that is faster to run, and can be a little smoother yet than Hybrid while still preserving good edge definition. SE also supports dual core machines, and some other benefits, with more to come over time to further differentiate it from Pro. However most people would do fine with Pro...

I spent the extra $$ for the studio version which uses their most advanced interpolation method (Hybrid SE). I am very pleased with the results. This sofware makes printing a joy to do.