- Written by Ofer LaOr
- Published on 18 December 2008
I will start by saying that I will have to review each part of the processing path separately. I think the DVDO “ABT102” (code name “Ducky Lucky”) algorithm is one of my favorites. It certainly ranks among the top 3 consumer de-interlacers that I have subjected to my torture tests. It rarely combs and squeezes out a lot more details than are available through other means.
The ABT1020 improves on this but only by a little. For my de-interlacing tests, I hooked up the same equipment to my LX5090 Pioneer FULL HD display (this is a similar display to the 5020 or Elite pro displays available in the US). The connections were made both directly to the display and also via the Edge.
My original display’s processing (note the stair stepping during motion)
DVDO Edge, the same program with similar motion (no visible combing or stairstepping)
Normal content had shown a dramatic improvement when running through the Edge. Differences such as de-interlacing artifacts, stair stepping and combing occurred on a regular basis when the screen had tried to deinterlace the image on its own. With lots of tweaks these can be reduced, but at the expense of a much softer image. The Edge retained the maximum resolution, but did not comb even once on my tests.
Scaling, however, was not at the top of the class. Some of the test patterns I ran through exhibited some ringing and were softer than I would have expected the image to be. There’s no doubt it’s a clear improvement over the Pioneer’s original processor, but further improvement can definitely be made.
I had also witnessed some white clipping above 95IRE on one of my analog PAL component sources, but the very same input worked great with RGBcvS.
The unit also did quite well with mosquito noise removal. The effect is subtle and can remove some image detail as well, but detail and edge enhancement can bring those details back. MPEG artifact removal is quite important when processing overly compressed bit starved sources and in particular – local SDTV channels.
When I compared this to the reigning champion of this field, the Algolith Mosquito, the latter performed better and removed more artifacts while retaining more detail. One algorithm that the Mosquito performs is Block and Macroblock artifact removal. This effect creates virtual blocks of similar, but not identical, shades and is one of the main causes behind making a channel appear “digital”. This effect is also part of the reason why gradients compress so badly – a gradient of blue sky usually deteriorates into a set of squarish blocks with varying shades of blue. Of course, the new H264 compression reduces these effects and can bypass some of these artifacts given enough bandwidth and time to do minor block corrections, but those of us who need these channels are stuck with blocks. The lack of a BAR algorithm is one that I hope will be addressed by ABT in the future.
I have already mentioned the detail and edge enhancement features. Both are similar in function and effect. Detail is primarily an algorithm that accentuates the area within a particular object’s boundaries, while edge is designed to make edges appear sharper. Both algorithms can introduce halos and add ringing to objects, but in small quantities, these can also counteract the softening effect of the mosquito algorithm and improve the details. These algorithms work quite well, and interestingly enough can also be set to negative values to deemphasize these very artifacts when they come from the source (many DVDs are notoriously plagued with over use of edge enhancement algorithms).
Aspect ratio controls are quite limited as compared to the VP series of products. In addition to underscan and zooming control, you only have 16:9, 4:3 and 4:3 LB (letterbox) on the remote itself. Add to that a new Panorama algorithm that is clearly a big step forward from all the other Anchor Bay processors. It’s still not perfect, but it is quite usable and is on equal footing or even better when compared to what most displays have to offer. As many displays block aspect ratio control when HD signals are sent to them, this is critical if you plan to use the unit with any sources that may not be 16:9.
HD is processed just fine, but not a significant leap from my own display’s capabilities
Many people ask about HD processing. If you have a full-hd display, which I would guess many of those reading probably do, you may be surprised to learn that your display probably doesn’t do too bad with 1080i content… Most displays are able to produce pretty good results. I have not seen dramatic improvement between native 1080i after processing it with this processor (or many others) although the change might be more apparent on a larger display (100” or so with a good projector). Content such as blu-ray usually goes through as 1080P anyway, so the main issue would be handling of 720P and 1080i from sources like OTA, satellite and cable channels. PReP also kicks in with those (including 720P) and picks up any 3:2 cadence it finds. In my testing, it also worked well with 2:2 content for 50 Hz content for the European crowds.
The unit lacks CMS and Gamma controls and is also not ISF oriented (no night & day modes). I doubt most users require this even from the higher end users, as these features would require some serious training or an ISF trained custom installer or calibrator to implement effectively.
As time progressed, the public beta of the DVDO Edge involved biweekly updates that often resolved issues and pushed the product further. The firmware update is one of the easiest I have seen to date. Turn the unit on with the “reset” button pushed in using a needle or ball point pen and connect the USB cable. The unit appears as a portable storage or thumbdrive and you can just remove the old firmware and put in a new one.