Importance of Optical Zoom

[HammerNoMore]

but many suggest it was folks confusing "raw" with "RAW" prefixing NASA-released images to the public

Leave it to the governement to confuse and ignore industry standards .

 

[archman]

Government? I figure it's all the fault of the programmers who designed the thing!

"Yeah guys, let's call it RAW, even though it isn't! We'll destroy civilization! Where's that new virus we were working on during lunch, anyways?"

 

[lukeROB]

I don't know what NASA's doing but a RAW image should be that, raw data, no compression. TIFF is actually the mother of all file formats. It's tag based and extensable. A TIFF file can be uncompressed, but usually it's LZW (the same compression used in GIF files). After JPEG came out they added tags for JPEG compression.

I know some of the camera manufactures take liberties with what they call "RAW" files but in actuality it shouldn't even have headers. If you want to load it into an imaging applications you should have to tell it the width and hight in pixels for it to be able to render it.

James

Hi James, I was debating this in another thread. Can you confirm that TIFF can be uncompressed in the sense you can have effective access to 'raw' information before any signal processing has been applied. Since most filters and processes can be undone through inverse procedures I always thought this would be possible. Also since noise is predictable for most processing algorithms this should also be able to be removed and the original pixel information restored. Am I wrong in thinking a totally lossless compression like tiff can have these attributres.

 

[ShoalDiverSA]

"ICER noticeably outperforms the JPEG image compressor used by the MPF mission and provides significantly more effective ossless compression than the Rice compressor used by that mission."

Looks like they applied the ICER compression to the article too...

Jewey4,

I have almost never had the need to use the zoom button on my camera underwater. You really need to get close to the subject before you hit the shutter release. There is a temptation when you first start out to "give it a shot" - zoom in, try to stop wobbling so much, and then snap it. The results are putrid - at least my attempts.

I would go for a camera that has the higher resolution (and manual white balance) at least. Unfortunately, the zoom figures for cameras have been so over-marketed with digital cameras, where the usefulness of a high zoom figure is probably limited in "normal" usage. You will need to evaluate what above-ground photography you will be doing. Will long zooms be required (wild animal or sports event photography)? If so, perhaps 10x zoom is justified.

4MP should suffice (it's what I have), but 5MP is useful for those photos where you didn't manage to frame the shot correctly, or where you would like to blow up a certain area for printing. You have the extra pixels to play with. If I want to make a print from 4MP, I pretty much have to leave the shot as is. Not much cropping can be done until I start losing resolution.

As far as the compression issue is concerned, always keep your original image. I edit my shots in Paintshop Pro, which works with the uncompressed image in RAM and saves compressed (JPEG) versions. So there is no image degradation in the session (this is pretty much the standard for most image editing programs). The real problem is when you go back and look at an image that you have clone-brushed the backscatter out of, have applied the Mandrake filter to, have sharpened and enahnced the contrast of and you don't like it. Then it is great to have the original image available... The raw formats are great for professional usage, but I really struggle to see image degradation on my JPEG photos (straight from the camera). After all, the JPEG format (as with MPEG encoding) is designed to be virtually unnoticeable to the human eye. A pixel by pixel comparison between raw and jpeg will reveal the differences. One thing to check is the level of compression that the photo performs on-board. This can be obvious to the naked eye, depending on the camera model.

Please note that I am only starting out with digital photography, so my opionion is not to be considered as gospel!

Cheers,

Andrew

 

[HammerNoMore]

Hi James, I was debating this in another thread. Can you confirm that TIFF can be uncompressed in the sense you can have effective access to 'raw' information before any signal processing has been applied.

Tiff I know pretty well. In the late 80's to mid 90's I worked for ZSoft (they were the first to do imaging software on the PC, in fact the paint program in Windows was one of their lower end applications bought by Microsoft). One of the things I wrote was the code to load and save TIFF files.

There are two different issue here. TIFF is a file format. As I said in a previous post it is proably the most extensive format available. Rather than being limited to a single compression algorithm like most formats, it uses a tag based system that lets the writer specify a host of different things, including which compression to use. The writer can also add custom tags for storing additional information. If these tags are usefull outside of your specific application they can be registered with Adobe and become part of the spec. This is how TIFF support JPEG compression even though TIFF predates JPEG by many years.

Currently TIFF supports the following compression algorithms:

Uncompressed
CCITT 1D (Compression used on fax machines)
Group 3 Fax ("")
Group 4 Fax ("")
LZW (Lempel-Ziv & Welch, compression used in GIF files)
JPEG (Compression used in JPEG files)
PackBits (An Apple Macintosh RLE compression)

The last 3 are color compressions and JPEG is the only one that is lossy.

So the first answer is that a TIFF file can contain data that is uncompressed, compressed with a lossless compression, or compressed with a lossy compression.

The second part of the sentence has to do with getting data before any processing is applied. Any processing that takes place on a camera is independant of the file format. One camera may do it and another may not.

Since most filters and processes can be undone through inverse procedures I always thought this would be possible. Also since noise is predictable for most processing algorithms this should also be able to be removed and the original pixel information restored.

Actually very few filters can be reversed. A simple filter like inversion (photo negative) can easly be reversed by re-applying the filter, but most others actually lose data. For example, if you brighten an image you are basicly adding a number to each color value. To make things simple I'll use gray data instead of color. Lets say you brighten by adding 10 to each pixel. 0 (black) becomes (10) dark gray. For this it is easily reversed as long as you know that the number added was 10. However all of the pixels that are in the range of 245-255 will become 255 (white) when 10 is added because 255 is te max. Now there is no way to determine what to do with those values. So essentially once you applied the filter you would have no black and if you tried to reverse it you would have no white.

I've written many filters and off the top of my head inversion is the only one I can think of that can be completely reversed.

James

 

[lukeROB]

Thanks, that explains it well. Stupid me - of course even a simple fliter like increasing values by 10 will put higlights out of range. There is also then the problem of an inverse filter not being able to deal with filter values close to 0 as 1/filter
(the inverse) it undefined at 0. So the raw data could be replicated within a degree, it could never be perfect. Highlights and shadows would suffer the most while midtones could be 'reversed' well. Anyway I get what you mean - thanks for the reply it helps a lot.

 

[HammerNoMore]

There is also then the problem of an inverse filter not being able to deal with filter values close to 0 as 1/filter

Actually an invert (photo negative) would be 255-value. Black 0, becomes white 255, etc. Like I mentioned one of the few completely reversible filters.

 

[lukeROB]

Actually an invert (photo negative) would be 255-value. Black 0, becomes white 255, etc. Like I mentioned one of the few completely reversible filters.

Sorry James, I was meaning the inverse in the sense of an opposite filter. Not neccessarily a invert (ie photo negative). Like taking any filter, say a blur, then 'inverting' (ie doing the exact opposite of the blur you applied. To get the orginal unblurred image - but I was basically stating that the like what you said before because of an limited pixel value workspace (0-255) there will be 'clipping' (not really the right word) when applying the blur and these clipped pixels will no longer be able to be reversed. I didn't really think this example through so a blur may not have this problem but I'm sure you get what I mean.

Again thanks for the help. Oh and again correct me if I'm wrong as I'm only just really learning and trying to understand this stuff. Cheers.