In camera processing RAW v JPG comparison

[Stephen]

Originally Posted by Saphire

I have just been running some tests on the above quote. Whenever I had to resize an image I would just resize in CS with image resize in small increments on a jpg image. I have just done a resize using the above method and the method I normaly use, I am very surprised at the diference at the same magnification. I have put a sample so you can see the diference.

Thanks Stephen for sharing that one I didn't know about it.

You are welcome

So, to sum up and clarify, are you saying that in your example, the left hand image in each pair is the one interpolated in the raw conversion, and the right hand image in each pair is using your regular step interpolation in CS? The difference is certainly marked. I wonder if the same difference is there with downsizing

FYI it is commonly accepted that stepped interpolation is not the best method, especially for relatively small upsizing. It is reckoned to be best to do it in one go using bicubic interpolation. Again I have never compared the two methods.

[Stephen]

Originally Posted by robski

LOL - Your up against a bunch of Real World engineers here who like to make improvements through science and measurement. ........................

LOL well you are all welcome to it I suppose someone has to do it though, and it can be interesting to see the results. The thought of photographing fonts and colour charts however is frankly a bit a bit of a *yawn

[Saphire]

Originally Posted by Stephen

You are welcome

So, to sum up and clarify, are you saying that in your example, the left hand image in each pair is the one interpolated in the raw conversion, and the right hand image in each pair is using your regular step interpolation in CS? The difference is certainly marked. I wonder if the same difference is there with downsizing

FYI it is commonly accepted that stepped interpolation is not the best method, especially for relatively small upsizing. It is reckoned to be best to do it in one go using bicubic interpolation. Again I have never compared the two methods.

Yes Stephen it is the left hand one that was done with your method and the right with just an image resize going up by a 100 pixels each time with bicubic until I reach the same size. I will edit the other to point out which is which.

[robski]

Originally Posted by Stephen

The thought of photographing fonts and colour charts however is frankly a bit a bit of a *yawn

The font and colour bars had nothing to do with photography. Look no camera hidden up my sleeves It was to illustrate how jpeg works as there are many misconceptions with regards to the subject. But I do take your point I'd much rather be out snapping away. Lets hope these dull gray days are soon behind us.

[hollis_f]

Originally Posted by Stephen

FYI it is commonly accepted that stepped interpolation is not the best method, especially for relatively small upsizing. It is reckoned to be best to do it in one go using bicubic interpolation. Again I have never compared the two methods.

Someone has though. After a similar discussion on DPReview Calvin Hass posted this comparison

[Don Hoey]

Raw shooter now installed but it may take a while for a result .......... 72 page manual

One benefit I suppose is that it forced me to do the SP4 upgrade and that took a while.

Don

[Don Hoey]

I have downloaded Rawshooters Essential to be able to look at the effects of converting an image from RAW using more than one converter.

Initial impressions of Rawshooter is that it is seriously fast in comparison to Nikon Capture v3. As mentioned in my previous post the manual is 72 pages long so a bit of playing is required. It has far more features than Nikon Capture v3 and appears to offer some of the features mentioned in Stephens earlier posts.

As I do not have many RAW images but I have found one that I has a range colours in a small area to show any effect on the image through RAW to JPG conversion.

NO PROCESSING OTHER THAN CONVERSION HAS TAKEN PLACE ON EITHER IMAGE.

The composite has received 15% compression to post it here.

Early days with this program but the results are way different.

Relative file sizes : Orig RAW 9.11mb Nikon Capture converted JPG 5.73mb Rawshooter converted JPG 3.46mb

Don

[Canis Vulpes]

I am not a fan of RSE and the above composite enforces my view. If you own a Nikon and shoot RAW NC is the quality way of converting RAW to Photoshop for further processing. NC is slow compared with others but worth the wait.

[Don Hoey]

Originally Posted by Stephen Fox

I am not a fan of RSE and the above composite enforces my view. If you own a Nikon and shoot RAW NC is the quality way of converting RAW to Photoshop for further processing. NC is slow compared with others but worth the wait.

From this comparison I would agree Foxy Bob.

A pertinant question may therefore be do say Canon, Pentax etc offer their own converter ? There seems to be more going on in the conversion than is immediately obvious.

Perhaps we need a techie to explain.

Don

[Stephen]

Well I'm sorry but I can't really see a great deal of value in Don's comparison. Please don't misunderstand me I have no loyalty to RSE whatsoever, but the same comments would apply whatever programs were used. How can you virtually dismiss RSE on the strength of a test which as I understand it has simply used the default setting within the two progs.

Surely a test would be better served if a file was adjusted to ones personal preferences. To me for example the RSE version is slightly warmer than NC. Maybe there is a tad less contrast. However its not my image and I don't have an idea in my head of what I would expect it to look like. Indeed another person may like the RSE version better, that would be down to personal likes and dislikes.

The point is though that had you used ACR in CS2 it would have applied different default settings and the image would no doubt look completely different. This default setting scenario is surely not the issue though. RSE like NC, C1, ACR, and Bibble all have the abilitiy to tweak the images ad infinitum to ones personal preferences.

I personally don't see how we can say from this example than one is better than the other. Just what would the criteria be that would say definitively that one is better than the other anyway

After looking at the two attachments again in PS I would also comment that the RSE version shows marginally more detail in the blacks, round the barge window for example. Also the full image has some serious compression artefacts, and neither of these are apparent in the browser, so where does that leave us?