So I have a problem where I have to severely crop an image to focus in on the subject. The image is sharp in detail out of the camera, however after I process the image and view it on a large screen (65") as full screen the image only takes up around 90% of the screen leaving a black background. This is not true of my laptop with a much smaller screen.
If I re-crop the original image to include much more of the shot, the problem on a large screen doesn’t exist - it takes up the full screen.
I’d be really grateful if someone could check the image attached and view it on a large screen, then offer some (if any) solutions.
I use a combination of RT and DT and used both Pix and geeqie to test with. If it’s helpful, the crop was done in DT v4.8.1.
Maybe the problem is just symptomatic of such a severe crop, and if that’s the case I’d love to know. If that is the case though, how would that affect the print quality at, say, A3 using a 20mb tif file? Would I just be wasting my time and money trying?
Nice capture, but you have so few pixels that it probably can fill your big screen at 100% magnification. It seems to me this image would benefit from upscaling the size. This is not something I normally do, but I am sure others can help here. Maybe read the DT user manual and reexport the image upscaled. darktable 4.9 user manual - export
You say that the image fills your laptop screen, but not your 65" (1.65m?) screen. That makes me think that your large screen has more pixels than your laptop (otherwise the image would fill both screens to the same extend); Note that physical size is in principle not relevant here.
When you allow upscaling in darktable export, that only does something if the requested output size is larger than the actual size of the image; and you’ll have to set an output size, if you set 0×0, there’s no reason to upscale…
Upscaling will (seem to) lose some sharpness when you watch the upscaled image at 100% on a small screen, but…
Such a large screen is not supposed to be watched from the same distance as your laptop screen, so a small loss of sharpness shouldn’t be noticable at a “normal” viewing distance (normal in this case being about 60", not 30 cm or about 6"). That’s why larger screens can still get away with the same pixel count as smaller screens (with the same aspect ratio…), see e.g. television sets.
That means that you don’t have to worry too much about the best upscaler while testing, I’d just start with darktable’s standard upscaling algorithms (they are there, and easy to use).
I used “set size - by scale (or file)” to 2 and that seems to have solved the problem, with no noticeable change in picture quality. the file size also jumped from 20mb to a more healthy 56mb tif.
Now in terms of print quality, would a 56mb tif print fine at A3 size, or am I likely to notice artefacts or blotchiness in such a printed image? Or would an upscale of 4 (rather than 2) be better for print?
The larger the print, the higher the typical viewing distance → the lower the print resolution (in DPI) can be. See, for example: DPI For Printing - Design Resources
Your upscaled file may be larger, but it is not detailed at all. Here’s a view at pixel level – you got ‘more pixels’, but ‘worse pixels’.
Compare that with the sharpness and noise level (or lack of the latter) seen in your signature – that is sharp at pixel-level, because it was created in high resolution, to start with):
Thanks for the friendly reminder, and your explanation.
Are you saying there is no point in upscaling, that because it’s such a severe crop that the print quality will reflect that? The image was shot in RAW, initially processed in RT then finished in DT, both using default DPI settings, which I would assume would be 300dpi for both tif and jpg? A cursory look in RT settings doesn’t appear to show anywhere to change dpi settings. Confused.
EDIT: This from Brave search…
"RawTherapee does not have a built-in option to change the output DPI (dots per inch) settings. By default, output files (such as JPEG or PNG) are saved at 72 DPI, regardless of the input file’s DPI.
This is because RawTherapee is primarily designed for raw image processing and does not natively support DPI settings for output files. The 72 DPI default is a common standard for digital images and is suitable for most purposes."
Doesn’t this seem counter intuitive if correct? Surely most users of RT would, by default need 300dpi for print? Surely it’s better to default to a higher dpi then downscale to 72dpi if required?
The DPI setting is just metadata.
If you know you want to print an A3 at 300 DPI (297 mm x 420 mm, or 11.7 inches x 16.5 inches), you need 300 * 11.7 x 300 * 16.5 pixels, so a resolution of 3510 x 4950.
If you want to print a billboard poster with an intended viewing distance of 3 metres, 60 DPI would suffice, so with the same number of pixels you’d be able to make a print that’s 5 times larger in both dimensions: 1.5 m x 2.1 m, or about 60" x 80", roughly 5’ x 7’.
DPI (dots per inch) is a way of mapping the pixel dimensions to physical dimensions. It’s simply a bit of metadata that says how many pixels map to one inch when printed. So if you have an image that is 1000x1000px, at 300 DPI the print size would be 3.3"x3.3", while at 72 DPI it would be 13.8"x13.8". The exact same number of ink “dots” would be used for both prints. So the resolution is the same, but the fidelity/sharpness would be different.
So I tried using the measurements you suggested (see pic), but the output tif actual measurement is 3464x4950. The pic itself looks fine. Will this pose a problem in print?
That matches the longer side of A3 alright. However, if your photo’s ratios (shorter side to longer side, in your case 1:1.429) don’t match those of the ‘A’ series of papers (1 to square root of 2, or 1:1.414), it would have to be distorted to fully fill the page. Is that what you want?
The dimensions you enter (whether in mm, inches or pixels) are bounding limits, the image will be exported to match them as closely as possible, without exceeding them.
This a whole lot more complex than i thought it would be. Might be partly why print is going out of fashion. I think I’m going to need some time to wrap my head around this.
So, what would I need to do while processing an image for it to come out exactly the size, for print, of an A3? IE: get both width and height the same. Or am I just confusing myself further?
You’d need to make sure you crop it to 1:1.414. You can enter that in the crop tool manually, and then save it as a preset, or add it via darktablerc - see:
But it is completely normal that printed material does not reflect the image 1:1, there is a finite precision when it comes to physical machines handling physical paper.
At 300 DPI, 3464x4950 pixels would give you 293.3 mm x 419.1 mm (so Adobe’s numbers were a bit off). As A3 is 297 mm x 420, this would mean an error of about 3.7 mm and 0.9 mm, respectively, or 1.2% on one side, 0.2% on the other.
Forget about the A3 paper for a second. Go to darkroom and after your cropping, look at the image information on the left hand side. The pixels shown in export width represent the available data/pixels you have. Then using that information you need to decide how many pixels per inch/mm are good enough for the size you want.
Maybe, for future reference, if such a situation arises again just forget about taking a pic that requires such a severe cut if the intent is for print. Instead invest in a lens which will get in closer without requiring a severe crop?
This was shot with a 55-250mm lens. Perhaps a lens up to 500mm might be better?
