Jump to content

andrew duncan

Members
  • Posts

    54
  • Joined

  • Last visited

Posts posted by andrew duncan

  1. Bob,

     

    If you can change over for $AUD 500, I think it would be worth it. Larger screen with more pixels, RGB histogram, Digic II processor, dust reduction (if it works?). From 6 MP to 10 MP is probably worth while too. Only down size I see is the batteries are different.

     

    Andrew

  2. Jonathan,

     

    I have no idea why there would be a difference with different orientations of the camera. The manual for the 20D (page 98) says

     

    "The EOS 20D can synchronize with compact, non-Canon flash units at 1/250 sec. or slower shutter speeds. With large studio flash, the sync speed is 1/125 sec. or slower..."

     

    Hope this helps,

     

    Andrew

  3. Sorry Anurag,

     

    What I was trying to say was that 'Lossless JPEG', for example JPEG-LS is another image compression standard. Don't worry about it!

     

    The problem with recompressing JPEG compressed images is that unless your program recompresses the image in the exact same way as the original, you will end up destroying more of the original data. This can be a real problem if you continually edit and resave your JPEG images. This is compounded by the problem that programs define their own quantization tables, which means that using the same level of comression in different programs produces different JPEG files. So, editing and recompressing in more than one program can cause image degradation.

     

    I was trying not to go into detail about JPEG compression, but it is hard to explain the problem without going into detail.

     

    If you have scans that were delivered as JPEGs and you want to rotate them then use Irfanviews lossless rotation plugin. Thats what I do!

  4. "What about the auto-rotate ..."

     

    No the auto-rotate features on digital cameras are just a exif tag. The lossless rotation in Irfanview is very different.

     

    The jpeg compression stream goes something like this:-

     

    1) Change colour space from RGB to YCbCr

     

    2) Chroma-subsampling (if applied).

     

    3) Discrete Cosine Transformation DCT

     

    4) Quantization (loss of precission) of DCT coeficiants

     

    5) Entropy coding (this is the actual compression).

     

    What appear to be a common misunderstanding is the the JPEG compression standard requires a number of steps (1-4) that make the image MORE compressible. Step 4 is where the precission is lost and where the lossy nature of the compression scheme comes from. Step 5 is the actual step that compresses the data using a lossless compression algorithm.

     

    The tranformation that Anurag was talking about decompress the image (i.e. the reverse of step 5), but work on the data that comes out of step 4. To view the image you need to do the reverse of all 5 steps, and this is what photoshop does when it opens a .jpg file.

     

    As for the help from adobe bridge, rotating an image 90, 180 or 270 degrees doesn't change image data (we are now talking about the uncompressed image in the application, not the compressed image in a file). While rotating at any other angle does change the data. I assume this is what they are trying to say?

     

    Andrew

  5. Anurag,

     

    <P>The JPEG rotation plugin of irfanview is derived from a program that is part of the Independent Jpeg Groups (<a href="http://www.ijg.org">ijg</a>) free library for JPEG image compression. It is loss less as the part of the JPEG compression that causes the loss is not recomputed. You cannot do the same thing in photoshop as after you rotate in photoshop the software needs to recompute all parts of the JPEG compression. There is another stand alone program called <a href="http://jpegclub.org/">jpegcrop</a> can do the same loss less tranforms available in Irfanview.

     

    <P>As an aside this is not lossless JPEG that is something <a href="http://www.jpeg.org/jpeg/jpegls.html">else</a>.

     

    <P>Hope this helps Andrew

  6. <i>The Vivitar that Alistair referred to will not work on any Canon EF mount since the Vivitar macro is an FD mount.</i>

    <p>I believe that the Vivitar 100mm f/3.5 macro lens is available in a number of autofocus mounts including Canon EF. I think this lens is made by Cosina and is the same lens sold by Phoenix.

  7. Serge, the quality factor is determined assuming that the jpeg file was created using the quantization tables used by the Idependent Jpeg Groups library (<a href="http://www.ijg.org/">IJG</a>). I am pretty sure that Nikon and Adobe use different quantization tables, so you can take the quality factor with a pinch of salt. However, for the highest quality you would expect a high value.

     

    <p>There are (at least) two ways of specifying the chroma subsampling. jpegdump reports it as horizontal and vertical sampling factors. Assuming that for id 2 and 3 the vertical and horizontal scaling factors are all 1, then Nikon also uses 4:2:2 chroma subsampling.

     

    <p>As you say, this is probably nothing to be concerned about on a camera that uses a Bayer pattern. That's what I was trying to say in my original post.

     

    <p>If you are looking for further reading, a good place to start is the <a href="http://www.faqs.org/faqs/jpeg-faq/">JPEG FAQ</a> and the <a href="http://www.faqs.org/faqs/compression-faq/">Compression FAQ</a>. They are fairly old, but contain some good information. I am sure there are other good references on the net.

     

    <p>Andrew

  8. Serge, chroma subsampling appears to kick in in PS at quality level 6 and is used at all quality levels bellow that. I use an old program called <a href="http://www.programmersheaven.com/zone10/cat453/17054.htm">jpegdump</a> to find out the level of chroma subsampling in the jpeg files from PS and the camera. I didn't discover this (although, I put 2 and 2 together about the JEIDA standard), I actually read it on another forum, and was surprised enough to confirm it for myself. I have some friends with cameras from other manufacturers and one day I'll get some jpegs from them.

     

    <p>Andrew

  9. One difference that doesn't get mentioned often is that the jpegs that come out of the camera (even when using the highest quality settings) have <A href="http://en.wikipedia.org/wiki/Chroma_subsampling">subsampled chroma</A>. If you convert a RAW file in PS or PSE and save a high quality jpeg chroma subsampling is not used. I have confirmed this with jpegs from a Canon 20D and I believe it is true of all the Canon DSLRs and many other cameras. It appears to come from camera manufacturers adhering to the 'Digital Still Camera Image File Format Standard' from <a href="http://it.jeita.or.jp/document/publica/standard/exif/english/jeida49e.htm">JEIDA<a>.

     

    <p>Having said all that the standard says to use 4:2:2 subsampling which means that two pixels share the same chroma values. That is, it is probably not a big deal as 2/3 of the colour information of each pixel is interpolated in most cameras anyway.

     

    <p>I also assume that this is why, if you use the black and white mode on the Canon 20D the jpeg files created are not greyscale jpegs, but full colour (YCbCr).

  10. Here is my favorite method of converting colour to black & white. It works well in Photoshop Elements

     

    Start with your colour image in PSE

     

    Create a Hue/Saturation Layer and change the blend mode to 'color'

     

    Create a second Hue/Saturation Layer and set saturation to -100.

     

    Now reopen the dialog for the first Hue/Saturation Layer and change the Hue and Saturation. You will get the same effect as using the Channel Mixer. You can refine the technique by only editing certain colours from the 'Edit:' drop down.

     

    Hope this helps

     

    Andrew

  11. Scott,

     

    Are they really TIFF files or (as Oscar has said) are they simply the CR2 files with the wrong extension. I would be very surprised if the Lexar Image Rescue software would change the image format.

  12. The COM marker is a part of the JFIF standard. The EXIF meta data is stored in an APP1 marker (this information is available from the EXIF 2.2 Specification in your link). As I understand it the APP markers were put into the standard to allow application specific data. I don't know much about IPTC, but I assume that it is also saved in an APP marker.

     

    The only problem with the COM marker is that (I believe) there can only be one COM marker in each file. So, to answer your question, the COM marker is idependent of the IPTC and EXIF information. The COM marker is a part of the Jpeg File Interchange Format (JFIF). To be pedantic JPEG is an image compression scheme, not a file format. The files that we all call JPEG files are correctly JFIF files.

  13. I live in Melbourne, Australia. I don't own a 30D, but I have had my hands on two that belong to friends of mine. One of my friends has reported some improvment in the autofocus from his 20D (sorry, I don't remember the exact details of the improvement).

     

    Hope this helps,

     

    Andrew

  14. When I bought the "Off-Camera Shoe Cord 2" I was surprised how short it was. It is probably more suited to using on a flash bracket (IMO), than anything else. I am currently exploring the idea of cutting mine and adding a short (1-2 meters) extension on it.
×
×
  • Create New...