Image Resolution...

[davegee]

The whole subject of Aspect Ratio / Image resolution / Zoom Degradation has provoked a lot of debate in various threads here over the last week or two. I have given my thoughts on Aspect Ratios and one possible way of approaching them. A question was posed regarding the correct/best Image Resolution and Zoom Ratio to prevent image degradation. The way I see it there is no single definitive answer because it depends on a number of factors:

Where you are starting from: Camera resolution?

Purpose of show: Your own monitor or everyone’s monitor?

For example: A D70 camera produces a 3008x2000 pixel file while a D200 file is 3872x2592 and the Canon 30D is 3504x2336. It is immediately obvious that zooming into a 1280 pixel wide image from any of these involves different percentages of zooming.

Also different monitor resolutions mean that the figure you are trying to achieve is different for each case. The “norm” these days is anywhere between 1024 pixels and 1600 pixels wide. I have attempted below, to produce a table which shows these differences. The figure given in each case is the Maximum Zoom possible from each of the camera files and for each monitor pixel size.

Definition: Maximum Zoom is the zoom amount beyond which some interpolation is required by PTE to fit the image to the monitor pixel size. It is assumed that some degradation of the image to some degree will take place beyond this maximum zoom figure.

Monitor Width: 1024 Pixels 1280 Pixels 1600 Pixels

3008x2000 (D70) 293% 235% 188%

3504x2336 (30D) 342% 273% 219%

3872x2336 (D200) 378% 302% 242%

Taking the full-size D70 image on a 1280 pixel wide monitor with a 3 to 2 aspect ratio and providing for the maximum 235% zoom would require a 3008x2000 pixel image which is at least a 1Mb JPEG file at compression 8. (The same figure for the D200 file would be 1.5Mb).

For PTE 4 the same full-size D70 image on a 1280 pixel wide monitor with a 3 to 2 aspect ratio would only require a 1280 pixel wide image whose file size at compression 8 would be just 215Kb. (The same figure for the D200 file would be the same – 215Kb).

The file size required to perform these extreme zooms is at least 4 times as big!!

The above table assumes that the show is being produced for use on the producer’s monitor. If the same show is uploaded to Beechbrook then the producer has to assume that someone, somewhere, is going to want to download and run on a 1600 pixel wide (or bigger) monitor. If this is accepted then the producer has to forget the figures for the 1024 and 1280 pixel wide monitors and work to the figures for the 1600 pixel wide monitor.

Let’s turn it around a little bit. If it is known that the zoom ratio required is going to be less than maximum, what effect does this have on pixel sizes and file sizes? The table below shows the pixel size and corresponding file sizes for D70 images which are going to be shown on a 1280 pixel wide monitor and subjected to the zoom percentages shown.

Zoom % 120% 150% 200% 235%

Pixel width

required 1536 1920 2560 3008

File Size 304Kb 456Kb 750Kb 1Mb

I think it is clear from this table that if keeping the file size of the final file to a minimum is important then tailoring the image size to suit the amount of zoom to be subjected to it will assist.

So what’s the conclusion?

The producer will have to prepare his/her own table of Maximum Zoom percentages based on his/her original image size and the purpose to which the final show is going to be put. A spreadsheet or table with all possible combinations would be a job for someone with much more time on his/her hands than I have.

DaveG

Sorry about the justification of the tables but I think it is still clear what the figures are.

[Barry Beckham]

Wow

What a lot of technical stuff..............but am I bothered

Arn't we in danger here of making a slide show appear far more complex than it really is.

Barry

[Lin Evans]

Wow

What a lot of technical stuff..............but am I bothered

Arn't we in danger here of making a slide show appear far more complex than it really is.

Barry

Hi Barry,

Indeed - but I suppose it depends on one's purpose for the show. From a photographer's perspective who produces images (and shows) for clients who are primarily interested in the aesthetics such as portraits, weddings, nature, landscapes, etc., I think the important issue is to keep the zooms to a reasonable amount and definitely concentrate on the artistic merits to evoke a feeling or "mood" for the production. In such cases it's only necessary to not exceed the practical limits and avoid creating too many artifacts of interpolation which could spoil an otherwise great production, especially when the show may be projected via an LCD projector, etc.

On the other hand, for the photographer who specializes in technical photography for a different client (circuit boards, highly detailed art objects for sale, etc.,) where there is sometimes a need for extreme zooms, the information is quite necessary and relevant.

But it gets even more technical - LOL Dave's approach assumes a linear relationship between pixel count and "resolution" which are actually "not" one and the same and his assumptions are based on primarily digital camera acquisitions. For example we would typically assume that two cameras producing equal file output sizes would have similar optical resolution and that we could then use the assumptions based on pixel count for the optimal zoom figures. In many cases this assumption is warranted, but certainly not in all. To be precise, we need to work from optical resolution rather than from native file size in pixel count.

As an example - my Nikon CP990 (3.34 megapixels native file size) produces an optical resolution of 900 horizontal by 900 vertical lines per image height while my Sigma SD10 (3.4 megapixel native file size) produces an optical resolution of 1550 horizontal by 1550 vertical lines per image height.

So we have two cameras with pixels counts which are practically the same, but which have vastly different optical resolutions. Not only are their true resolutions extremely different, but the degrees to which the images can withstand interpolation and still maintain excellent image quality differ even more. My 3.4 megapixel SD10 actually outperforms my eight megapixel Canon dSLR's and my 10 megapixel Sony R1 when it comes to interpolation and holding image quality. In fact, it approaches my 11 megapixel Canon 1DS (4064x2705) in enlargeability for some types of capture and exceeds it for others.

Then there is the entire area of scanned film which is another complete can of worms because the true "optical resolution" and thus ability to withstand interpolation may indeed be much higher than the pixel count from the scan would imply.

The bottom line is that we can only "approximate" these figures and must take each individual photograph as a unique part of the equation. We simply can't create a "rule" but rather a "guide" for this type operation. As long as we realize that we are dealing with "a rule of thumb" we won't get into too much trouble, I think.

But Dave's point is well taken - we must indeed be sensitive to the degree that we incorporate zoom and the eventual environment where our shows could be displayed. Of course we must first and foremost be attuned to the aesthetics and not loose the "baby" in the bathwater......

Best regards,

Lin

[nickles]

But it gets even more technical - LOL Dave's approach assumes a linear relationship between pixel count and "resolution" which are actually "not" one and the same .....

Hello Lin,

I was very fascinated with your response. Much insight on your part. I think I learned quite a bit from your narrative....it does get a litlle more complicated when you view from the technical vantage point...especially from the perspective of optical resolution.

Thanks for sharing.

I also found Dave's explaination with regards to zooming and required image sizing very insightful and well composed also.

Thank you also for sharing, Dave.

Aren't we in danger here of making a slide show appear far more complex than it really is.

Barry

Barry, I also appreciate and understand where your statement is coming from. Sometimes we tend to make things much more complicated than they ought to be. But, I do think people need to be aware that if they over-zoom that they can cause a good image to appear as a lousy image.

Sincerely,

ken

[Barry Beckham]

Well I think that should be common sense for anyone photographer or not, but then again some of the demos I have seen, obviously not.

However, it may be simpler to say that, if you intend to pan or zoom an image the picture needs to be bigger that the rest to allow the quality to be retained during the movement or zoom.

Barry

[LumenLux]

For such a disagreeable subject, there seems to be a lot of agreement here!

Thank you Dave, your table is useful and very much like what I was suggesting someone could do for us all. Your explanation is also meaningful. And Lin, you have answered some of the questions I had not yet formulated. And Ken and Al, thank you for saving some of us from personally experiencing complete brain drain.

My own, irreverent and non relevant conclusion - reminds me of a book title that read:

"Everything in this book is false - but that 's the way it is."

And yes Barry, we are all still having fun with it. Thanks for reminding us.

[nickles]

"Everything in this book is false - but that 's the way it is."

Lumenlux,

Could I borrow that book when you're through with it?

Ken

[davegee]

Thanks Barry,

Quote:

"However, it may be simpler to say that, if you intend to pan or zoom an image the picture needs to be bigger that the rest to allow the quality to be retained during the movement or zoom".

I merely quantified, to the best of my ability, what you choose to generalise on.

DaveG