Is an 8x10 a humongous poster? 'cause it takes 29mp to print
Leaving aside some commercial options, I think it is worth mentioning that HP/Canon/Epson home printers print in dots-per-inch of C, Y, M or K -- not Pixels-per-inch. A 24-bit "full color" pixel on screen is not translated completely into one of those 600 or 720 (or 2400) DPI on paper. Even the nicer 6, 8, and 12-color printing systems are still "dots of component color" -- not PPI in the on-screen sense.
From what viewing distance?
Yes, I know. When I said...
I meant 600/720 pixels from the image per inch.
That is not correct. A complete translation to 600 or 720 PPI does occur. Qimage refers to these resolutions as the “native” resolution of the printer (I call them the photo resolutions.) Modern Canon printers will print at 2400 DPI vertically and between 2400 and 9600 DPI horizontally using drop sizes of 1 and 5 picoliters. This provides matrix sizes of between 16 color dots (4x4) and 64 color dots (16x4) to describe a single color pixel from an image at 600 PPI. I don't use Epson printers, but the information I've read indicates that it also uses 16 color dots to describe a single pixel, using an 8x2 matrix.
I'm just wondering how much of that difference is autosuggestion. For more general results a double blind experiment should be done. For personal use such results as yours may of course still be useful
That is certainly possible, although far from proven.
The ability to perceive life-like motion and the the connection of frame rate to it is relatively well known fact. However, it is mostly a separate subject from the resolving ability of the human vision.
As most serious amateurs and FPS pros do. Nothing new here. With modern graphics cards you usually can increase the resolution to 800x600 or even higher though when playing older games, since in most cases the frame rate at low resolution is not limited by the graphics card but rather by the processor or memory subsystem.
Oy...maybe when I was a kid. Righit now anything closer than 8 inches starts getting blurry.
That's true, but the noise from a sensor is a lot closer to mathematically-generated noise than you think, in character; you seem to be getting your idea of what a sensor generates from conversions . Conversions filter out noise near the nyquist through demosaicing and NR. All converters use significant NR even when you turn NR "Off" - all that "off" means is that the least amount of NR that the converter wants you to see is used. All RAW data is much noisier than you seem to think.
Until recently nobody had to pay attention to downsampling algorithms for DSLRs, the required native resolutions on printers 300-600 PPI and 360-720 PPI usually ask for upsampling. To avoid downsampling one can set the printer driver on the higher print resolutions (dpi), better weaving and the requested (input) native resolution will be 600 or 720 PPI. A 10 x 7 inch print from an A900 24 MP file isn't downsampled (or upsampled) when 600 PPI input resolution is requested by the print driver. That will not be the real optical resolution available in the print but it will not downsample the image before it is processed in the printer driver and if noise should become visible in a print one has to aim for a print quality like that at that size or print 4x larger in area at a lower quality setting.
Down-sizing the JPEG gave worse results with nearest-neighbor and better results with Gaussian. But the camera creates the low-res JPEG from a full-res RAW, so it's not really a good comparison.
Well, that appears reasonable to me. Nevertheless, if you want to compare the image quality, you have to compare images of equal size. This may be difficult and there may be obstacles. You may choose to upsample up to 10 or 12 MP or downsample to 6MP, whatever, but you just can't post crops of two images of completely different size next to each other and assess the comparative image quality based on these crops. But that's what is being done here and it's just wrong.
This is so trivial, I don't think he would even care to discuss this.
When most people say "reduce noise" they mean "increase SNR".
Yes, moving away from a source is equivalent to downsampling it.
--
