ON1 Resize 2023 - Wrong workflow for printing?

[Dennis]

Advertisement (gone after registration)

Hello everyone

I downloaded the AI software a few days ago to resize some 24MP M10 files for printing. Still on free trial time. For example, a 5911 x 3941 file has the maximum print size (as per file) of 19.7" x 13.1", of course, at 300 dpi. But if I want the same file to be printed at 24" x 36" without resizing, the dpi will be 164, which is not enough if the view distance is close. I tried the ON1 plug-in and standalone app to resize to the wished largest size, and I guess the new file is outstanding, without any artifacts. It's like magic. But I still have one significant doubt.

I understood that the best practice is to edit (in my case in Lr Classic) the file, and when everything it's done, as the final step, apply the resize. Am I right? But from which file type? Because I know the software can create another file, such as PSD, DNG, or TIF, so I can export the final resized file for printing directly from LR. But I could also export a 300 DPI, a not compressed Tif file from LR at 16bits (for example), then launch the ON1 Resize app and convert the TIF file to a new resized Tif file (or high-quality Jpeg)... What do you recommend doing? With the latter, am I losing quality in the process? Because it looks like the options are more than one, but the best quality is (maybe) assured with one type only of workflow. Please help me to understand,

Thank you,

D.

 

 

[dpitt]

Maybe some of the pro's will correct me, but I think you are over complicating things. Even the 10 MP files from my M8 are good enough for any print size if you view the prints from a normal distance. I had them printed on 70cmx50cm without noticeable issues. Your 24 MP could go to 40" with the same resolution.

For prints of 36", I think 160 DPI is plenty, one should never look at this from closer than a meter and even when you go up to 30 cm (1 foot) you can not see pixels from a dye sublimation print.

The main quality factor will remain your lens and focusing. It needs to be spot on for these enlargements.

Just my 2 cents...

[Advertisement]

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

[Dennis]

9 hours ago, dpitt said:

Maybe some of the pro's will correct me, but I think you are over complicating things. Even the 10 MP files from my M8 are good enough for any print size if you view the prints from a normal distance. I had them printed on 70cmx50cm without noticeable issues. Your 24 MP could go to 40" with the same resolution.

For prints of 36", I think 160 DPI is plenty, one should never look at this from closer than a meter and even when you go up to 30 cm (1 foot) you can not see pixels from a dye sublimation print.

The main quality factor will remain your lens and focusing. It needs to be spot on for these enlargements.

Just my 2 cents...

Thank you @dpittfor your comment. I see.

I have read and studied much about this, but I still have some doubts. It looks like there are two types of thinking about the topic 🙂 The ones (like you) who believe that 160 DPI is fair enough from a decent view distance (VD). Then people think that without considering the VD are happy to embrace new technology and make significant prints even better. Maybe a few years ago, when the only tool available was Photoshop, the result of an upscale image was uncertain. But now, with fantastic results using software like ON1 or Gigapixel, I think that taking advantage of this can (maybe) give us a better print... even from a closer VD.

Also, I'm talking about fine art prints: rag cotton, bamboo, or alpha-cellulose. Printed large and sold not cheap for clients. So, if the picture looks fantastic not only from far away but even from very close distances, it's a win-win... 

Lastly, my question about the workflow is also because I'm considering buying an M4/3 camera, such as an OM-1. And If I also need to crop a bit, the file size is not that big like the M10 if I want a 24" x 36" large print.

Still studying and understanding. I guess I will need to enlarge first a few large prints to see how big I can print w/o resizing and to know if I'm happy with the result.

[dpitt]

4 hours ago, Dennis said:

....

Lastly, my question about the workflow is also because I'm considering buying an M4/3 camera, such as an OM-1. And If I also need to crop a bit, the file size is not that big like the M10 if I want a 24" x 36" large print.

...

It is extreme, I admit it. In my experience inherent quality of the shot is the most important factor. AI can maybe help up to some point. If you would live nearby I would show you my 30cmx40cm prints of my Digilux2 (5 MP) , it has a crop factor of 4. You would not know if I did not tell you.  Superior lens and perfect matching sensor prevail. There was no up sampling involved from my side. I do not know how the lab treated it, but it was a cheap print in B&W.

If you see these, you will not be afraid of going to M4/3 with less pixels. Of course you do not have the luxury to crop very much. If I calculate correctly your 36" would require about 20MP to get the same resolution as my Digilux 2 prints. And that does not take into account that you won't look at them at the same distance. I think you could get away with 10MP.

[Dennis]

5 hours ago, dpitt said:

It is extreme, I admit it. In my experience inherent quality of the shot is the most important factor. AI can maybe help up to some point. If you would live nearby I would show you my 30cmx40cm prints of my Digilux2 (5 MP) , it has a crop factor of 4. You would not know if I did not tell you.  Superior lens and perfect matching sensor prevail. There was no up sampling involved from my side. I do not know how the lab treated it, but it was a cheap print in B&W.

If you see these, you will not be afraid of going to M4/3 with less pixels. Of course you do not have the luxury to crop very much. If I calculate correctly your 36" would require about 20MP to get the same resolution as my Digilux 2 prints. And that does not take into account that you won't look at them at the same distance. I think you could get away with 10MP.

Thank you for the invitation. I guess I'm not nearby to enjoy your Digilux print, but If I have a chance, I'll let you know 🙂 Appreciate! 

Meanwhile, cause I never printed fine art bigger than 12" x 18", I decided to place an order next week: two "test" 24" x 36" prints for two BW photos: once from a slightly cropped M10 24MP + an uncropped one from my Q2M.

FWIK, an 8368x5584 Q2M file, can be printed at 300 DPI up to 28" x 19" (as per file). So, I guess these tests will give me a better idea of what a print will look like at a specific/given distance with both resolutions. If I'm delighted with both results, I won't need any AI resizing software. Does it make sense? 

Thank you!

 

[Dennis]

6 hours ago, dpitt said:

If you see these, you will not be afraid of going to M4/3 with less pixels.

It will be interesting to see large prints quality and differences of

• OM-1 (micro 4/3 20MP) resized at 40MP
• OM-1 with High Res shot (50MP handheld)
• Fuji X-H2 (APS-C 40MP)
• Leica M10R (40MP)

Because at the same 40MP, the sensor size is also making a huge difference.

[dpitt]

Advertisement (gone after registration)

20 hours ago, Dennis said:

It will be interesting to see large prints quality and differences of

• OM-1 (micro 4/3 20MP) resized at 40MP
• OM-1 with High Res shot (50MP handheld)
• Fuji X-H2 (APS-C 40MP)
• Leica M10R (40MP)

Because at the same 40MP, the sensor size is also making a huge difference.

The quality of the pixels is of course important. FF tends to win over cropped sensors. But if the FF is not as good and the 4/3 is very good it has a chance. I just meant that the amount of pixels is secondary to quality of lens and sensor.

I would be concerned performing tests like this without tripod. It will add a variable and possibly skew the results of the test. Printing good large prints is one thing, shooting good high resolution shots handheld that hold up in large prints is an other.

[dpitt]

21 hours ago, Dennis said:

FWIK, an 8368x5584 Q2M file, can be printed at 300 DPI up to 28" x 19" (as per file). So, I guess these tests will give me a better idea of what a print will look like at a specific/given distance with both resolutions. If I'm delighted with both results, I won't need any AI resizing software. Does it make sense? 

Thank you!

 

Indeed. Compare the cropped M10 file printed at the same size. It will have less than 300 DPI, and then see if it makes a huge difference for you compared to the Q2 print of 300 dpi

[hansvons]

On 2/10/2023 at 5:31 PM, Dennis said:

What do you recommend doing? With the latter, am I losing quality in the process? Because it looks like the options are more than one, but the best quality is (maybe) assured with one type only of workflow.

@Dennis , I can relate to your thinking as I also went through that.

From the technical side, the more information you have from the onset, the better your result will be. That's why there's Tiff (uncompressed and in 16-bit - never 8-bit) and JPEG, which is 8-bit and compressed. The latter is only usable when the file's size/weight matters as a delivery. But as you want to print, the print is you're deliverable; hence, I'd sent the printshop a 16-bit Tiff. That also counts for the postproduction work you do when roundtripping applications. In your case, that would be LR - 16-bit Tiff - scaling programme. 

AI scaling can do astonishing things as it adds to the regular scaling algorithms such as linear, Lanczos, nearest etc... machine learning and, thus, can mitigate their particular shortcomings. One of the significant goals of scaling is to maintain sharpness without introducing artefacts. If you sharpen your image before you scale, you will make the scaling algorithms' life much harder. The best practice is to do the sharpening after the scaling, as sharpening depends very much on the pixel size, which will change inevitably when you scale your image to another size.

The same can be said about de-noising. De-noising softens pixels. Often, it's better to de-noise just the chroma and keep the natural texture.

I cannot comment on ON1, but I find Topaz Gigapixel an excellent sharpening tool when scaling 1:1, meaning the software is only "improving" the image without scaling it to a new size. 

I don't print myself as I know this is a science of its own, and I'd need more time to become a proper printer (and likely miss the talent). I find convincing contrast (especially micro contrast) and sharp pixels/texture without artefacts more important than the sheer resolution. That's why the lens is so critical, as micro contrast and sharpness are determined by the lens. The less you tinker with sharpness in post, the better your result will be.

I shoot at 24 MP and do occasional prints. I don't mind 300 dpi vs 160 dpi, as these are only numbers. The proof is in the pudding/print.

 

Edited February 14, 2023 by hansvons

[hansvons]

On 2/11/2023 at 5:13 PM, Dennis said:

Lastly, my question about the workflow is also because I'm considering buying an M4/3 camera, such as an OM-1.

Are you considering purchasing the OM-1 for wildlife photography? This is a bit off-topic. I recently purchased the OM-1 with the 300mm f4 lens and the 1,4 teleconverter. I do occasionally shoot birds and insects, but only privately.

That combination is a winner in many regards, as it's super-handy, weighs almost half of a comparable FF rig (had before a Z6 and the 100-500) and has the best-ever stabilisation. The bird AF algorithm is excellent; the ProCapture allows you to go back a second before pressing the trigger. The latter feature alone is worth the purchase.

The sensor and colour are suitable, albeit not Leica (in my case, the SL2S). But I don't plan to photograph people with the OM-1, so I don't care about skin tone reproduction. That little camera is a brilliant niche system.

How it will print, I will be sure to see. The noise is alright looking once the noise colour is suppressed. Again, the lens is the starting point. For wildlife, it cannot be sharp enough. The 300mm f4 is sharp, very sharp.

[Stuart Richardson]

In my experience an M10 with a good lens will get you to about 60x90cm before you really start to feel a softening. I agree with the general consensus that the overall technical quality level in the whole imaging chain is critical for large prints. You need to have a stable platform or very high shutter speed, optimal aperture (typically not wide open or f16 etc, unless those are what you need for the photo), as low an ISO as is possible (or better stated, as optimal an ISO as possible in cases of dual ISO and other more tricky situations), a good sensor and good processing. I do not typically use AI resizing algorithms, as in my experience they introduce an artificial look in the images when inspecting closely. Often you will have very defined edges, but then unnaturally smooth areas in between and so on. I am sure they are getting better every day, however. I have not tried ON1, but I would recommend you go over the image with a fine toothed comb to see if there is anything unnatural.

The best advice I can give as a printer is that if you are going to be printing big, you should be aware that what you see at 100% is what you will see in the print. If it bothers you on the screen at 100%, it will bother you in a large print. I agree with Hans about DPI, however. DPI is not a hard and fast rule, and the requirements are going to be wildly different depending on the subject matter. A portrait without a lot of fine detail will likely look great even at a low DPI, while a very detailed landscape or similar image might fall apart at 150dpi. Typically, however, in medium to large prints I think it is pretty easy to get away with 150dpi or so, though it is of course better to have more. As you are no doubt aware, megapixels are not created equal. At the time when I had both the M10 and the Panasonic S1, it was clear that the Panasonic S1 has a much sharper sensor. Despite both being 24mp, shots taken with the same m lenses (the longer ones not affected by smearing) were significantly sharper on the S1 than they were on the M10. Trust your eyes. Unless your eyes are bad, haha. Then trust mine, lol.

[Dennis]

6 hours ago, hansvons said:

From the technical side, the more information you have from the onset, the better your result will be.

6 hours ago, hansvons said:

One of the significant goals of scaling is to maintain sharpness without introducing artefacts.

6 hours ago, hansvons said:

If you sharpen your image before you scale, you will make the scaling algorithms' life much harder. The best practice is to do the sharpening after the scaling, as sharpening depends very much on the pixel size, which will change inevitably when you scale your image to another size.

6 hours ago, hansvons said:

I find convincing contrast (especially micro contrast) and sharp pixels/texture without artefacts more important than the sheer resolution. That's why the lens is so critical, as micro contrast and sharpness are determined by the lens. The less you tinker with sharpness in post, the better your result will be.

Thank you! I understand that having an excellent file to start with and knowing it well as it could behave in prints it's crucial.
It makes sense to do the sharpening after scaling (if needed). My initial question of the thread is: do everything necessary in LR, then upscaling + sharpening? Or, as you are saying, even if not upscaling, the sharpening with Gigapixel that I heard is probably the best. Downloaded already. We'll see.

It's also important to judge our files objectively and see their differences. It should apply here also that the practice and the experience make the master 🙂

5 hours ago, hansvons said:

Are you considering purchasing the OM-1 for wildlife photography?

Actually not. It would be used as a daily street/landscape/macro camera. And as a travel tool when needed. For its small size and weight, huge DoF, macro ability, and advanced IS (also for the high-res handheld), it could be a great companion for my M10 and Q2M for my daily walks looking for a great shot to print and sell. I'm starting a new print business, and for this, I want to produce a good and varied body of work and series on a weekly basis. For the M43, I was thinking either w/ the 12-100/4 (for flexibility) or 12-40/2.8 Pro II (smaller, lighter and extra stop) as first lens ... I still can't decide.
You should print a large print with your OM-1 and let me know 🙂 

 

[Dennis]

3 hours ago, Stuart Richardson said:

In my experience an M10 with a good lens will get you to about 60x90cm before you really start to feel a softening.

60x90 is about 24"x36", which is the size I picked for the M10 print. It will be the most significant print of my M system. Hopefully, on Friday, I'll receive it. I can't wait to understand better about this 🙂

3 hours ago, Stuart Richardson said:

I do not typically use AI resizing algorithms, as in my experience they introduce an artificial look in the images when inspecting closely. Often you will have very defined edges, but then unnaturally smooth areas in between and so on.

I just started to deal with AI software at the beginning of February. Sometimes I don't see anything terrible added to the image; sometimes, there are indeed artifacts, and it looks so fake or so digital. I guess. As you say, I need to trust my eyes and see carefully what is natural and smooth and what is not.

3 hours ago, Stuart Richardson said:

The best advice I can give as a printer is that if you are going to be printing big, you should be aware that what you see at 100% is what you will see in the print. If it bothers you on the screen at 100%, it will bother you in a large print.

This is actually and very simple but effective advice that totally makes sense. Thank you

[Dennis]

This is the picture I sent to print in 24" x 36" .... 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited February 14, 2023 by Dennis

[Dennis]

@hansvons- @Stuart Richardson

I received yesterday the large prints. OMG, I wasn't expecting such excellent quality on 24" x 36".
The Q2M file comes from another planet, lol. I could see super details from even 30cm, which was remarkable. While the slightly cropped M10 file was, of course, not as good as the Q2M, but still very impressive. So considering the recommended 6ft+ view distance, the print result is just outstanding. Not sure why, but somehow I was expecting less quality. But not for the resize software marketing but for what most people are saying about the topic "How big you can print." 
And I discovered (also thanks to you) that having an excellent file to work with,  I mean exposure, details, micro contrast, etc., it's crucial.

Enjoy your weekends,

D.

 

[Stuart Richardson]

I remember having that experience when studying printing back in the day at ICP. I had a class with Brian Young who is a master printer that worked for places like Magnum, and people like Bruce Davidson and James Nachtwey. He talked about how a certain kind of photographer would say that you cannot print 35mm any larger than 11x14, because according to them, bigger than that and the print would be too soft or grainy. He pulled out a 30x40 inch darkroom print of one of James Nachtwey's photos he took in Afghanistan (before 9/11). It was shot on tri-x and 35mm and it looked incredible. These were analog prints, but the same wisdom applies. He basically demonstrated that pronouncements about image quality, DPI and so on are often formed out of ignorance or an overemphasis of the technical over the aesthetic. I pride myself on being a skilled technician, but I also have printed enough beautiful artworks that were deeply technically flawed to know that it is really the aesthetic impact that is the most important. Building a feeling for what is acceptable is part of being an artist, and it is something that you only really gain through experience. Of course, the technical plays an important role, but it is only one part of the equation. I am glad that you were happy with your prints!

[dpitt]

5 hours ago, Dennis said:

@hansvons- @Stuart Richardson
And I discovered (also thanks to you) that having an excellent file to work with,  I mean exposure, details, micro contrast, etc., it's crucial.

Exactly. Welcome to the club.

[Dennis]

11 hours ago, Stuart Richardson said:

He basically demonstrated that pronouncements about image quality, DPI and so on are often formed out of ignorance or an overemphasis of the technical over the aesthetic.

It sounds wonderful. Just like that!

It happens in many ambits, but it's the truth. I enjoy playing with the technical side of things, but I don't appreciate being overwhelmed. I want the flow while I'm taking photos. Sometimes I don't shoot for a week, sometimes every day, and it's okay. But every time it happens, I enjoy the process a lot, and usually, I'm satisfied with the result, at least one keeper, or many, who knows. A good harvest is always welcome 🙂 I use the technical part to help me achieve my goals, and I spend time studying gear before buying. But once I have it, after a (hopefully short) trial time, I prefer to use it to my advantage rather than taking part in affecting the shooting process. 

7 hours ago, dpitt said:

Welcome to the club.

Thank you! 

.

The OM-1 ... mmm... now that I saw other 12x18 prints (by a Nikon D5), despite the different sensor sizes, they are 20MP. 

As the second lens, the new 90/3.5 macro is very tenting. For the first one, I'm between the

• 12-45/4 
• 12-40/2.8,
• and sometimes the 12-100/4, but I have not decided yet. For street, landscape, urban architecture, macro, travel... Which one do you recommend? For lightness, and size I would say The first one. SUper light and small, and i will shoot at f4 or f8 maximum any way...

Don't worry; in my thread there is no off-topic 

Edited February 19, 2023 by Dennis