I have a basic question and I am embarrased that I should know the answer. But I want to be certain!! If I shoot the D800 with the options for NEF and JPEG, is the JPEG image downsampled when it goes on the SD card? I know that downsampling is resizing the image to a smaller size...and I know that saving an image in JPEG is saving it to a smaller (compressed) size. SO... is a JPEG image on the D800 SD card already a downsampled image when I copy it to my PC, and therefore I need not do anything further in order to benefit (noise reduction) from downsampling?

Dan

"My most rewarding photos are those that capture something I didn't 'see' in the frame....so just SHOOT."

Visit my Nikonians gallery.

Don't confuse image size with file size. Both the nef and jpg shot in fine, large size have the same number of pixels and the photo dimensions will be the same. The only time the camera down sizes is when you shoot the photo in medium or small size. Then the pixel dimensions change.

Another example of compression would be an audio file comparing a full size wav file with an mp3. The length of the song doesn't change, only the quality of the audio. The same comparison is made with a nef versus a jpg.

Len

Visit my Nikonians gallery.

So...when advisors say the "when you shoot FX and it is downsampled from 36mp to 16mp it reduces the noise", what do I need to do? When I shoot in FX how do I get it downsampled?? I know what you mean by not confusing file size to image size but I am still not clear how I get a downsampled image.

Dan

"My most rewarding photos are those that capture something I didn't 'see' in the frame....so just SHOOT."

Visit my Nikonians gallery.

You downsize the image in your editing software.

In PhotoShop CS5 for instance, when you open the photo one of the items in the options bar at the top of the page is "Image". If you click "Image" a drop down menu appears with one of the options "Image Size". If you click "Image Size" you will get a box marked Pixel Dimensions which will show the Width and length of the image in pixels. These dimensions are linked by default so if you change the width from say 7,360 pixels to 3680 pixels and then click "OK", you have downsampled the image. The new image size is now 3680 x 2456 pixels.

This is how I downsample my images when posting a photo on the web like in a Nikonian's post. I change the width to 1000 pixels which is a nice size for viewing.

edit. I see you posted a D800 image in your gallery. The photo has been downsampled (downsized). How did you do that?

Len

Visit my Nikonians gallery.

Thanks a bunch..

See the post in response to my question from Eric...below. You both sound like you know what your talking about and this downsampling is different for me because I always shot in JPEG. I know there is compression but the results in JPEG (only once saved) were typically good enough the way I handled the copies and for WEB albums. Now with the 800 I want to maximize resolution and make bigger prints so I guess I need to shoot in NEF or RAW. Thanks for the explaination above. I knew some of that but your explaination was worded more clearly than my thoughts. (usually the case lately).

PS: Actually I did not resize the picture on my gallery....the site did that automatically. I just joined not too long ago and am still learning the tricks.

You guys (and gals) are really helpful here!!!

Dan

"My most rewarding photos are those that capture something I didn't 'see' in the frame....so just SHOOT."

Visit my Nikonians gallery.

You don't need to "do" anything.

I'm going to overly simplify this and hope to hit the main concept.

The D800 image measures 7,360 x 4,912 pixels. At 300 ppi that translates into an image that is 24.5 inches x 16.4 inches.

The D4 image measures 4,928 × 3,280 pixels. At 300 ppi that translates into an image that is 16.4 inches x 10.9 inches.

But if you want to produce two prints of the same size at 300 ppi, you need to reduce the D800 image to a file that is 4928 x 3280, so for nearly every 2 pixels on the D800, you need to convert them to a single pixel. That means the software has to resample the image and determine the "correct" value.

If one of those 2 pixels is noise, and the other is not noise, the blending of the two pixels will effectively make the noise go away. It's the same idea when you take a full size image and resize it for posting on the web.

Now if you have to take a small file and create a large print, the opposite occurs. For every noise pixel on the D4 sized sensor, you will convert it to 2 noisy pixels at a D800 resolution.

Eric Bowles
Nikonians Team
My Gallery
Workshops

Nikonians membership — my most important photographic investment, after the camera

Thanks Eric,

Let's see if I have it right.

I should shoot in FX mode on the D800 and record it NEF or TIFF. Then I put it on my PC via SD card and downsample it... lets say to half size. Then save it as a TIFF. Now my image has been downsampled to 18mp. Is that right?

If so... to maximize resolution, I should never record in-camera in JPEG. Always use TIFF or NEF. Copy from the SD card to the PC and do the size reduction. Is that right?

Additionally...I should not record in-camera to JPEG format unless I do not want maximum resolution. Is that right?

Finally...JPEG format images are always reduced in resolution due to compression. Is that right?

PS: thanks (in advance)for the course in Downsampling 101!!

Dan

"My most rewarding photos are those that capture something I didn't 'see' in the frame....so just SHOOT."

Visit my Nikonians gallery.

I don't think you normally need to vary your workflow to emphasize or de-emphasize downsampling. It happens automatically. From a theoretical standpoint, you might get slightly better results if you resized every image to the size of your final output. In practice, when you make it smaller the image is already rendered with appropriate resampling to fit the output.

To test your logic, substitute the "resize" for "downsample". All you are doing with downsampling is resizing to a smaller size and using a resampling algorithim to determine the resulting image. Your file format does not make much difference (although you might have small differences in the related editing capability).

So the answer is yes - if you resize and image to a smaller size it will be downsampled. This is going to be the case for JPEG, TIFF or NEF files.

I don't see any problem with using JPEG's straight from the camera if the situation calls for it. Furthermore, you can specify the appropriate size for your JPEG in the camera. If you use a JPEG, it will still be downsampled if you want to print it as an 8x10 or some size below the full resolution.

As Bill suggests, you won't necessarily eliminate all noise, but you will in most cases reduce it's impact.


Eric Bowles
Nikonians Team
My Gallery
Workshops

Nikonians membership — my most important photographic investment, after the camera

Eric,

If one of those 2 pixels is noise, and the other is not noise, the blending of the two pixels will effectively make the noise go away.

Both of the photosites have signal and noise.
When you combine them the signal gets added. s = (s₁ + s₂) / 2
When you combine them the squares of the noise get added together. n = sqrt((n₁² + n₂²) / 2)
So the Signal to Noise Ratio (SNR) get higher and the noise is less apparent (it does not "go away").
(s/n is higher than both s₁/n₁ and s₂/n₂)

Regards,
Bill

Visit me at My site

Programs line genuinefractals (or I think it is called Perfect Resize) from onOne, describe how they do a better job than PS for upsizing. I assume there is no advantage for using something other that PS for downsampling ?