June 13, 2020
For some time I have been thinking about just how much resolution is needed to satisfy my most extreme needs for photographic images. Todays cameras offer not only excellent resolution (45-60Mpixels) but also excellent dynamic range and noise performance. This combination exceeds even the finest film performance for 35mm and, in some cases medium format. But is it enough?
Up to this point my philsophy has been to develop the discipline to create the highest quality "digital negative” consistent with the quality of the camera, lens and the shooting environment. Whether the image is to be posted on Instagram or rendered on to a large print, I want the choice to repurpose my best images and that demands the source negative be as good as possible. Fortunately, image editing software technology is improving rapidly so older images can see the benefit of new software. Thus, over time, older digital negatives can be rendered to higher quality based on the capabilities of new software. This is particularly the case for noise reduction, sharpening and image up-sizing.
I continue to believe that the ultimate expression of a photographic image is in a print and I have worked to master printing techniques which give the highest quality, reproduced in a consistent and predictable manner.
Today, I am constrained by the size of print I can make at home. I have an Epson 3880 which can print up to 17” X 37.4”. Red River Paper makes panorama paper at 13" X 38" (330mm X 965mm) which also prints on the 3880. Using a RIP, larger sizes are possible. The native resolution of the 3880 is 360dpi although it prints just fine at 300dpi
An uncropped image from the Nikon Z7 (8256 X 5504) will print 23” X 15.3” at 360dpi. Big, but not that big.
So, what would be the maximum practical size? How about 100” X 67”? That’s a very large print and I can’t imagine needing or wanting anything larger. I’d want the print to have resolution that I could get close and see a lot of detail. I don’t think it needs 360dpi but it probably can’t be less that 150dpi. So, that’s the target.
This type of goal could easily argue for medium format cameras. I’m going to exclude this and stick with full-frame. Most of my work involves moving around (travel) and medium format is just too bulky. Further the gap between full-frame and medium format is diminishing over time. Also, the lens choice for full frame gives much more flexibility
What are the factors that contribute to optimizing the quality of the digital negative? Here are a few:
1. Use the best camera you have in terms of resolution and dynamic range - ex. Nikon Z7, D850 or equiv. and make sure you are familiar with the controls. Shoot RAW, avoid lossy compression. Use the best quality (prime) lenses.
2. Make sure the camera settings are appropriate for the type of shoot. Make sure the shutter speed is sufficient to prevent motion blur. Make sure the ISO is at the lowest level possible. The aperture should be set to enclose the DOF required for sharpness. If any of these conflict, consider using a tripod and remote release or take control of lighting
Note: typical rules of thumb for DOF and motion blur tend not to be adequate for today’s high resolution cameras. This is a separate topic but need to be determined for the specific camera you are using.
3. Consider using ETTR techniques to maximize dynamic range giving flexibility in post.
4. If necessary, pay attention to the natural light. If it’s not right, consider waiting until things change. Or move. Be patient.
5. If the subject requires quick response, consider pre-focussing and other set-up actions to save time. Be ready
6. Be aware of the need to crop. Try to avoid as much cropping as possible. Move closer to the subject if practical. Or use a bigger lens
7. Use the correct focusing techique for the type of subject. Incorrect focussing will destroy resolution. Be aware of things that can cause mis-focusing - focus shift, field curvature etc. and mitigate if possible.
What are the factors which can destroy the (technical) quality of the image or prevent it’s use in large prints?
1. Cropping. A 50% crop (not untypical for wildlife) will reduce resolution (and print size) by a factor of 4. Try to get as close to the subject as needed to minimize cropping. Pay attention to composition when your taking the picture instead of correcting in post.
2. ISO. Increasing ISO reduces both dynamic range and SNR. (put data in here). Most modern cameras handle noise very well and there are excellent noise reduction apps out there. However, there’s no substitue for a “clean” noisless image giving the full dynamic range available from the image sensor. If this is a limitation, consider using faster lenses which let in more light (at the expense of DOF). Or light the scene.
3. Mis-focussing. Anything that prevents you from getting the sharpest possible image is an issue. For focus this could be a number of factors:
(i) Focus variability - most camera / lens combinations will show variability on focus from shot to shot. Take multiple shots
(ii) Calibrate the lens / camera combination for optimal focus (not necessary for mirrorless cameras)
(iii) If you’re manually focussing, use magnified view and focus peaking. Consider using a loupe.
(iv) if possible, focus at the target aperture. Some lense exhibit focus shift across the aperture range. Most cameras do AF wide open so it’s possible to mis-focus while the image in the viewfinder looks sharp. Note, this is a complex topic, see diglloyd.com.
(v) Unanticipated motion blur. For wildlife photography I need at least 1/2000 sec shutter speed to freeze motion. High resolution cameras have much smaller pixel pitch so even tiny movements can cause motion blur when viewed at 100%. For me, most images which look out of focus, are actually caused by small motion blur.
(vi) It’s obvious, but make sure the exposure is correct. Blown highlights can destroy an image. On the other hand, underexposing is as bad as using high ISO (sometimes worse).
Ok, now that you have taken the “perfect” picture, what options are there to achieve the maximum print size with the highest possible quality?
The baseline is the “normal” post workflow - image enhancement, noise reduction, sharpening, cropping etc. I won’t cover this here since there’s plenty of other places to get guidanc on this process. The big question is what do you do if you need more (print) resolution than is provided by the camera alone.
Here, the options are:
1. Up-scaling. This is the process of increasing the pixel dimensions of the image. This is accomplished by using interpolation algorithms to “create” inter-pixel values that approximate what would be the actual value if the image were larger. There’s plenty of software out there that will accomplish this - Photoshop, Gigapixel AI, ON1 Resize, etc. Each has it’s own pros and cons. Right now, Gigapixel AI seems to provide a superior solution. There is a further process which takes multiple images, slightly offset and computes sub-pixel values which are more accurate than normal interpolation. It’s arguable if this process produces superior results but is worth a try (get a link).
As an example, take a typical Z7 image (cropped square) and up-scale by 2X - 4946 X 4946 -> 9892 X 9892. At 150dpi, this could be printed to the following dimensions: 66” X 66”. Here’s an example of this up-scaling using Gigapixel AI:
Full image @ 9892 X 9892:
8” X 8” crop @150DPI (not to scale)
You can see that there’s plenty of resolution to go around.
I tried a number of other images, including landscapes and achieved mixed results. My conclusions on up-scaling are as follows:
1. GigaPixel AI does a pretty good job at up-sclaing, at least to 2X in each dimension (generally better than PS). With today’s cameras, this will yield images which meet or exceed the print size objectives I’m looking for so long as 150dpi is acceptable.
2. I realized you need the very best lenses and technique to maximize the result. Small amounts for mis-focus, motion blur or lens softness just get multiplied up when you up-scale. Same goes for noise. I tried up-scaling some images taken with the NIKKOR 24-70 f/4 S on my Z7. While this is a nice lens, it just didn’t give me the sharpness needed to see the full benefits of up-scaling.
When using flash or studio lighting, motion blur is minimized due to the fast speed of the flashes, which helps. Also, for lit portraits, using something smaller than wide-open aperture reduces the risk of mis-focusing. The image above was taken with the Nikon 105mm f/1.4 portrait lens (natural light @ f/1.4) which is a good example of “best in class”. For landscapes, using a tripod and remote release is recommended. If you need sharp focus across the entire image, consider focus stacking.
3. Post processing is a challenge. In general, it looked like post processing on the final up-scaled image yielded the best results but this is not necessarily case for all images. Plus, these large images take their toll on the processing power of the computer. So experimentation is required.
4. The nice thing about up-scaling (relative to other methods like stiching panoramas) is that you can capture movement. The images don’t need to be static. Any method which requires taking multiple images in sequence and blending, or stitching, requires completely static subjects or you will get “weird” results.
More options:
2. Create a panorama.
Placeholder: Panorama Rig: https://petapixel.com/2017/03/28/built-panoramic-photo-rig-made-6-nikon-dslrs-awesome/