exporting to SEP in 8bit or 16bit

[stump4545]

Advertisement (gone after registration)

if i am not intending to print my images, is there much lost in exporting to silver effects in 8 bit rather then 16bit?

 

for black and white images or color images same thing?

 

 

 

 

 

just that 16bit really slows down my imac compared to speedy 8 bit.

 

thanks for the help.

[ghoonk]

Doesn't that really depend on how you're presenting your images?

 

With 8-bit, you get 254 shades of grey between white and black (total 256)

 

With 16-bit depth, you get 65,534 shades of grey between white and black (total 65,536)

 

The M9/M9P/MM shoots in 14-bit, so that's over 16,000 shades per color channel, and though some people argue that it's not true 14-bit (more like 12-bit), that's quite a bit of data to throw away.

 

Of course, I could be wrong...

[Ecar]

No - no big difference in real life.

Yes, you are throwing out quite a bit of data, but if your pictures are meant to be seen on a screen you'll be compressing the files anyway and the color/grey rendition will be much more affected by the viewer's equipment and its calibration than by the fact that you worked in 8bits rather than 16 bits...

In any event, why don't you just process an 8bit version and a 16bit version of the same image and decide by yourself whether 16bits are worth the trouble?

[stump4545]

what do most people here post process with do 16 or 8?

 

i also heard that the 16 bits only has a real effects if the image is to be printed.

 

for viewing on a monitor no difference 8 or 16bit. is this true?

[adan]

1. When possible, it never hurts to work with the greatest bit depth possible. It leaves overhead for adjustments and manipulation without "combing" the data (winding up with a histogram with a lot of spikes and gaps).

 

2. But anyone who thinks any current reproduction method - even printing - delivers anything even remotely resembling 65,534 grays - or 65,534^3 colors - should try the math.

 

Let's assume a good printer can actually output 360 pixels per inch. A top-end Epson at finest dot size lays down 2880 dots per inch. That means it prints an 8x8 array of dots for each pixel - or 64 "levels" from no dots (white) to pure black (64). With a black ink and two gray inks, that gives 64 x 3 = 192 grays. Although that assumes the printer uses "light light gray" to modify even the darkest near-blacks, which I doubt actually happens. But we'll be liberal in assumptions. Also, at 2880 the printer can print two dot sizes, so we get 384 gray levels from three inks.

 

Strictly in monochrome, we'd need about 170 different gray inks to really reproduce 65,534 distinct grays. At least three times THAT to do color at "16-bit".

 

Laserphoto printers can do better, since they can flicker the laser for more density/exposure variation within each pixel. Even so, the Lightjet is only 12-bit, and Fuji Frontier printers are 8-bit.

 

The good news is that human vision is limited to just under 24-bit (8-bit x 3 colors) - we can distinguish about 10,000,000 colors, whereas 24-bit color can reproduce 16,777,216 colors. Which is already overkill, except that it allows for differences in gamut (a given 24-bit color space may be reproducing reds we can't see anyway, but just barely including greens and blues that we CAN discern, or vice versa).

 

Attached is an 8-bit gray square. Somewhere within it is a shape in a slightly different gray (1 gray level lighter or darker, out of 256 levels). Can anyone visually distinguish the shape?

 

It is there (despite jpg compression) - you can download the image and run a threshold tone curve to find it - but your eye will never see the difference even on lowly 8-bit.

 

3. Still, 16-bit is not a bad idea when possible, for the reason mentioned in (1), and because printing/screen technologies may improve in the future. After all, a decade ago we only had 4 inks (CMYK) to work with, instead of 8 or so today.

[stump4545]

thanks for explaining.