That would be the sharpness setting on the PDX-10. I put it, and all the other custom presets, in the default position by performing a reset. The default is in the middle of the scale. Minus 3 clicks would turn it to the minimum. The VX-2000 was set the same, however even with matching settings less edge enhancement is visible there.

Some picture contents (leafs, rocks...) "look" sharper when there is some edge enhancement. During standard shopping demo's an
attracive element for many people... and smart sales people.

Funny you should mention that Andre.... yesterday evening I took my cameras out back and filmed a few seconds at each of the sharpness settings... in a scene with rocks and leaves!

Results to follow.

Boyd, it was quite a long swim for observing you shooting...But back to reality...what looks awfull when applying too much(and thick) linear edge enhancement are human faces at quite some distance, like shooting groups of people.

For your consideration, a few more tests inspired by this topic:

http://www.greenmist.com/dv/garden Here I photographed an outdoor scene that has a lot of texture and fine detail using my still and video cameras. As I did with the resolution charts, I created a reference image by shooting a 2560x1920 high quality JPEG with the Nikon 5700 which I downsampled to 720x480. Quite a bit of detail is apparent here. Once again, I then compressed this JPEG with the DV coded, and the results may surprise you; it is very hard to tell any difference. I repeated this test several times, and actually compressed it into a new Quicktime DV movie which I used for my frame grab to insure that it really was DV compressed.

This image looks considerably better than either of the video camera images. So yet again I must conclude that the DV codec gets unfairly blamed for creating blocky images with lots of artifacts. I think that this test, and the resolution chart tests I shot before, support the thesis that the real culprit is our (relatively) cheap DV cameras! I would be interested to see any other tests that lead to different conclusions. But at least I now understand why the high quality broadcast cameras cost so much more than the small chip prosumer cameras we all know and love :-)

And once again, to put this all in perspective, I am looking at a small section of the image that's been magnified. The overall video frames look pretty good by themselves.

You will also find a variety of other new tests here that I've just completed, including the effect of the sharpness custom preset: http://www.greenmist.com/dv. Enough testing for now... phew... time to get outside and enjoy what's left of this beautiful holiday weekend!

Awesome work Boyd! Thank You for all that food for thought.

Congratulations Boyd. These are unique pics for those "who want to know..and see". Maybe one point again: uprezzing in the (VX2K) cam generates a lot of artifacts (vertical straddling) mainly because of the limited rescaler performance. A better result ( even better than the FCP upez) is to go straight into the display scaler with the 4:3 pic. Especially modern high-end displays can do an allmost perfect uprez job. See http://users.pandora.be/andredeclercq/

That is very interesting Andre. I assume the "cinema 16:9" was shot on the VX-2000 as letterboxed 4:3? It does look a lot cleaner. I will have to take my letterbox shot, apply some black bars and feed it to my Sony 16:9 screen to how its scaler compares to the FCP stretch, which I felt was disappointing.

Also try this: take my VX-2000 image that was shot in 4:3 and framed for 16:9 http://www.greenmist.com/dv/16x9/07.JPG and put it into photoshop. Now crop it to the top and bottom of the chart and you should end up with 720x360. Finally, resize to 854x480. The result is substantially better than any of the other VX-2000 widescreen examples I posted. I should probably include this on my webpage.... ah, and I thought I was done! So I'd imagine that something like After Effects might be the way to go if you're shooting widescreen on the VX-2000. I don't have that program so I can't try, but DVFilm Maker does this I think, so maybe I'll run the image through that.

Fascinating topic, there are a lot of turns in this road...

I don't know the exact concept used for of the VX2K uprez, but I think they use a fixed polyphase filter technique with linear interpollation, and because there is no framebuffer it has to be applied on field basis which makes linear interpollation more prone for the vertical artifacts we see. But even if the framebuffer would be there (and smart deinterlacing and reinterlacing would be applied) it would still be inferior (I think) against a performant direct uprez into a higher resolution (768 pix in my 37"LCD TV)
And indeed rescaling a Photoshop still frame should be allmost perfect because of the more sophisticated interpollation used for PS. But unfortunately this algorithm would be difficult to get it real time for video images, and moreover this ideal uprezzed pics would need to be rescaled again in most 16:9 displays.