TEST FRAME SHOOTOUT
The Test Frame Shootout pits Futureware's Format Conversion against
Photoshop's interpolation techniques, Qimage's Lanzcos and Vector techniques, and LizardTech's Fractal interpolation
technique. The input in all cases is a single frame of 360X288 PAL video, and is
converted to a frame of 1440X1080 HD. The ideal screen resolution for the
Shootout is 1600X1200. The input frame is shown below.
| Photoshop Nearest Neighbor |
A simple unsophisticated technique that shows generally poor
results, although color fidelity is exact. |
 |
| Photoshop Bi-Linear |
A more complex scheme, tends to blur fine details. |
 |
| Photoshop Bi-Cubic |
A more complex curve-fitting algorithm, shows fairly good
results, although color aberrations are significant. |
 |
| Qimage Lanzcos |
Slight quality improvement over bi-cubic. |
 |
| Qimage Vector |
An intelligent algorithm that gives visually pleasing,
although inaccurate, results (note color aberrations, for
example). |
 |
| LizardTech Fractal |
Considered the premium of all curve-fitting algorithms. |
 |
| Futureware
Markov-Predictive |
The statistical algorithm gives good clarity, best
fidelity and excellent anti-aliasing. |
 |
| Futureware
Markov-Predictive |
The statistical algorithm gives best results in all areas (text, clarity,
fidelity) when applied to motion video. |
 |
|
