Updated January 5, 2008
Updated August 10, 2008
Reshade Client-Side Beta Evaluation
Update: Reshade Image Enlarger is out of beta
see the latest version of the photo enlargement software here.
I’ll describe in a little more detail some of the controls available in the new Reshade client-side version. Here’s how the options bar appears:
The main controls are:
- zoom
- image size
- edge control
- smooth defects
- contrast details
- accuracy
- edge threshold
- best quality
Zooming, with the zoom slider, or entering a value in the zoom text box offers an easy way to view the image at different sizes, enlarge the image and see a quick preview. If you require more control use the image size box where you can enter the new width and height of the resized image.
The most important option available is the Edge control. This allows to adjust the sharpening of the resized image. Expect significant changes to the resulting photo enlargement when varying this. Generally an optimum value is used when the auto option is checked.
Several other options complement Edge control: Contrast details, Accuracy and Edge threshold. Contrast details permits a different kind of sharpening that works by accentuating contours giving them a more 3D look. Accuracy influences the way edges are approximated in the resized image. Less accuracy usually means rounder edges while more accuracy means higher precision and, on very large zooms, a finer artistic touch. Edge threshold makes possible selective edge transformations. Lower edge threshold allows more edges to be transformed. Higher threshold only transforms stronger (less smoother) edges.
Smoothing can be used on the image to remove noise and image defects while doing the image resizing. It’s very helpful combined with higher values for contrast details.
Finally, even better zooming is possible enabling the best quality check box.
Check out the photo enlarger and see how you can order it here.
Updated August 10, 2008
How Much Money is in Photo Upscaling
A quick definition of stock photography might be helpful here. The wikipedia entry says that it consists of existing photographs that can be licensed for specific uses. Book publishers, specialty publishers, magazines, advertising agencies, filmmakers, web designers, graphic artists, interior decor firms, corporate creative groups, and others use stock photography to fulfill the needs of their creative assignments.
There are many big stock photography sites online. A quick search on google returns popular names like corbis, fotosearch, comstock. The businesses prosper by selling photos at different sizes, as the client specifies. The prices for these sizes vary to suit everyones needs: for small images the price is low, but these can mainly be used only on websites and other mediums that don’t require great resolution. The expensive large sizes are suited for printing, displays, posters etc.
We’ll talk more about this price - size dependence. Here’s a plot depicting this:
The blue line represents a plot of the price of an image found on stock photography sites (average price for the mentioned sites) versus the image size (in megabytes). The red line represents the best linear fit. This gives a trend for estimating larger size prices. The megapixels of the photo can be calculated by dividing the size in megabytes by 3.
A more detailed view of the data used
The resolution of the photos can be derived from their size (in MB) like this:
640 KB - approx. 640 x 480 pixels; 8.9″ x 6.7″ at 72 ppi
2 MB - approx. 1024 x 1280 pixels; 14.2″ x 17.8″ at 72 ppi
14 MB - approx. 1700 x 2550 pixels; 5.7″ x 8.5″ at 300 ppi
32 MB - approx. 2800 x 4200 pixels; 9.3″ x 14″ at 300 ppi
50 MB - approx 3400 x 5100 pixels; 11″ x 17″ at 300 dpi
Using the linear estimator from the figure we can calculate the price for a 150% and 200% times larger image. These factors are considered in the zoom column from the table below. The megabytes have a quadratic dependence on the zoom factor: Let’s take the 200% zoom (from 8 to 16, respectively from 60MB to 240MB). This means that the new size will be 200% the width * 200% the height of the photo. This gives a 4 times larger photo (60*4 = 240 MB).
| zoom | MB | price |
| 1 | 1 | 211 |
| 8 | 60 | 476 |
| 12 | 135 | 813 |
| 16 | 240 | 1284 |
We get that by doubling the size of the image we triple the price. It amounts to a 1284-476 = 808$ difference in price. The sites don’t offer these large sizes. It’s not that they are not needed, they actually are. But there are no such powerful sensors (cameras or scanners). It would require a 240/3 = 80 megapixel sensor. Professional cameras go as far as 20 megapixels. That doesn’t mean there are no higher resolution images. Several photos can be combined in panoramas and such.
So let’s just suppose that the linear trend holds. If upscaling photos and maintaining just 66% of the original quality this means doubling the price of the image (3 * 0.66) and a 400$ difference. So, doubling the size of a good quality image, of about 20 megapixels costs about 400$.
We’ve considered the worst case scenario, and still there is great value added by upscaling photos. The photo enlargement algorithm here at reshade.com makes possible the above mentioned quality upscaling and on most images it does even better than considered (a lot better than 66% of the original quality). Resize an image online for free and see why: online image resizer.


