Video Game Processing

A few specific comments...

You can "chain" Pelican system selectors to get as many switched inputs as you need. Pick a video processor on processing--not number of inputs. My setup has a component signal passing through three system selectors, with no significant visible degradation.

You'll definitely want to find an XRGB-2 or XRGB-2+ on ebay since you have all those old consoles. The XRGB-2(+) can convert 240p signals (composite, S-video, and component) to 480p VGA out. You can use the XRGB-2(+) before another video processor, or attached directly to the monitor.

The VP-50 can process 240p signals properly, but it only does a good job of this on the component inputs. Or, perhaps, maybe it's the devices I'm testing (NES, SNES) that are putting out a poor 240p signal. In any case, the XRGB-2(+) processes 240p signals from old consoles very well. The VP-50 does a better job than the XRGB-2(+) on component 240p signals (for playing PS1 games on PS2, or emulated old games on new systems).

The VP-20/30 can likely also handle 240p signals, but I don't have one to test to confirm this. If it's possible, you would need the ABT-102 add-in card. Maybe somebody else can test this. (Check on "Game Mode 1"--it only works in that deinterlacing mode.) Even on the VP50, proper 240p handling is an unadvertised feature, and nobody from DVDO spoke up to point out that it had it, even when people were specifically requesting it (not aware that it already had it).

Pay attention to the processing delay on various processors. The VP-20/30/50 have low latency modes with lower quality deinterlacing, but the VP50 is actually worse on progressive inputs, until the promised firmware update to add the option of low latency on progressive signals is released.

Yesterday I swapped my Crystalio 1 for a VP50 unit - with mixed results concerning videogame processing.

I still generally like the SD processing of the Crystalio (Faroudja DCDi-based) better than the VP50's. The multidimensional sharpening and the Detail Enhancement of the FLI-23xx does quite a lot for SD material, especially for videogames. The VP50 is - I admit - close, but can't quite match it. The VP50pro with added Fine Detail Boost will probably do the trick.

The downside of the C1 was that the lag was quite high (about 60ms), which is of course better with the VP50 with it's two game modes.

Since I play a lot of shoot'em ups 240p handling was essential for me. The C1 did this very good. It recognized 240p properly and the interpolation during the scaling only produced a few "shuddering" pixels here and there. Otherwise very nice though.

My "test game" always was ESPGaluda. It's a Cave Shoot'em up with lets you switch between 480i rescaled graphics and pure 240p during gameplay. Right in the first level you'll find aircrafts which throw a shadow. To imitate transparency the shadows are flickering (in240p), in other words: they are only displayed every 2nd frame. A typical effect on classic raster-based videogames.

The C1 did an incredible job on processing this effect with no artifacts at all present. Unfortunately the VP50 *FAILS* in Video mode and Game Mode 2. The rest of the picture looks nice, but those shadows get ugly as hell with flickering on the outside and massive combing errors on the inside. Only Game Mode 1 processes 240p properly, but you're definitly loosing on detail enhancement and picture calmness. Especially the edges of any text have much less definition and detail on the Game Mode 1.

All in all, the results are pleasing though the C1 with lower latency would rock at videogame processing. It's a pity that 240p is mistreated in Game Mode 2 and I hope ABT gets back to tweak this mode on day in the future.

If anyone is interested in pictures I can try to take snapshot of the faulty deinterlacing.

Tobias

I'd love to see pictures of the comparison--if you have the time!

I can't take pics of the old setup, since I have disconnected the C1 already.

Below you'll find a few comparions between the VP50's Game Mode 1 and 2. The game is ESPGaluda in 240p. It's a scrolling shoot'em up, so the shots are a bit blurry, since the game gives a black screen once paused. Always take a look at the shadows produced by the aircrafts.

Game Mode 1 = low resolution, "bobbing" effect (shuddering lines), proper 240p handling:

http://pms.hazard-city.de/mode1a.jpg
http://pms.hazard-city.de/mode1b.jpg
http://pms.hazard-city.de/mode1c.jpg
http://pms.hazard-city.de/mode1d.jpg

Game Mode 2 = better resolution, no "bobbing" effect, but improper 240p handling:

http://pms.hazard-city.de/mode2a.jpg
http://pms.hazard-city.de/mode2b.jpg
http://pms.hazard-city.de/mode2c.jpg
http://pms.hazard-city.de/mode2d.jpg
http://pms.hazard-city.de/mode2e.jpg
http://pms.hazard-city.de/mode2f.jpg


Tobias

thanks for the pics. Makes it pretty clear which one works and which one doesn't! I noticed you're not running your dreamcast through VGA. Any reason?

??? ESPGaluda is a PS2 Game. My hookup is actually:

DVD Player (SDI) -> VP50 (SDI)
LD Player (HLD-X9) -> VP50 (S-Video)
Digital Receiver (RGBS) -> VP50 (RGBS)

PS3 (HDMI) -> VP50 (HDMI)

PS2 (YUV) -> XSELECT-D4 (YUV) -> VP50 (YUV)
XBox (YUV progressive) -> XSELECT-D4 (YUVpro.) -> VP50 (YUV)
DC (VGA) -> XSELECT-D4 (VGA) -> VP50 (YUV)

The XSelect-D4 is a RGB/VGA/YUV Switch with integrated color transcoder, so the DC signal gets transcoded to YUV progressive before it gets scaled. I could use the RGBHV input on the VP50 as well, but I didn't have the cable on hand, so I wired it the way above for now.


UPDATE: just connected the DC directly to the RGBHV input of the VP50. While the quality is like expected, the voltage level of the VGA Box (original Sega) seems to be too high. The DC bootscreen (with the DC whirl logo) doesn't appear white, but dark grey with wrong colors. The DC menu and the actual game is ok though.

I know--I should have hit return--those two ideas weren't related.

I've been looking for a VGA cable for my dreamcast(s) for quite a while. eBay is ridiculous, and racketboy--HOLD ON!!, now it IS in stock! I'll be buying that after this entry...

Anyhow, I was interested in a lumagen HDQ and wanted to hook the DC's VGA up to its RGBHV, so I'm quite interested in your impressions of that hook-up vs. the XSelect-D4. I'm guessing it rocks and rocks hard. It's got to beat (my) s-video.

I'm impressed at your system list! I thought I was bad...

Do you mind if I ask you what kind of equipment rack/shelving you use with that many systems? I've procured 2 8ft. server racks for that purpose but need to purchase or make shelving within. It was the only way I could think of organizing them all without simply stacking them to the ceiling. If this is too far off the original topic, I don't mind responses in a PM...

I'm a bit confused about your comment regarding voltage. It looks incorrect, but not during gameplay?

And one last "more-on-topic" question if you have time--do you bother calibrating your systems output(s)...and how? Any test patterns hidden in games in the NES era onward?

clarified s-video statement

It's good to hear that the Crystallio 1 handles 240p correctly--this feature isn't given much attention so it's good to know another device that handles it.

One comment on your analysis--I don't think it matters that any other mode on the VP50 doesn't support Game Mode 1. 240p handling should be a "perfect" process without any heuristic guessing necessary (like is necessary for 480i deinterlacing, where missing information is being generated). Extra latency would not "assist" in the process at all. To the extent that there is shuddering in Game Mode 1 on 240p content (what you describe as a "bobbing" effect), I believe is actually slight noise in the signal which causes the diagonal interpolation to "jump" back and forth when the decision on which direction to interpolate is right on a threshold. This is only a guess, and I don't have proof of this, and it would be difficult (although possible) to construct a test case for this, without access to the code. If I am correct as to the cause, then this can be addressed by changing the diagonal interpolation algorithm to not use hard thresholds.

The only reason that 240p handling should be enabled on other modes is so that the feature is easier to discover. (If the other modes supported it, the performance would be the same regardless of what mode was selected--except that when a different mode was selected, there would be additional latency needlessly.)

no idea why I missed your last posting back in late august.

The thing the VP20/30/50 does wrong in Game Mode 1 is the following: on a 240p signal all 60 fields are to be displayed on the same height level. This is why a classic CRT TVs give you the typical scanlines.

What the VP20/30/50/50pro should do is this: it should take every of the 60 fields without moving it up/down compared to the previous field, double it to 480p by adding the same field ONE LINE down. and then scale to the output resolution. This would display single horizontal lines rock solid - just like on a CRT.

What the VP20/30/50/50pro DOES INSTEAD is this: it moves every 2nd field down one line before doubling and scaling it. (this is done to keep the original field order and position of the fields on 480i material). This makes the screen a bit shaky, giving what I described as the "bobbing" effect on my posts before.

The problem MIGHT be that the DVDO VPs can't tell a 240p signal from a 480i signal and thus can't decide wether to keep the fields on the same height level (for 240p) or adding an offset to every second field (for 480i material).

This could VERY EASILY be changed via a manual switch in the options though. Let's keep the fingers crossed that it gets added one day.

Tobias

Sorry for the huge delay in replying. Been extremely busy lately.

I believe the VP50 does keep every scanline on the same height with 240p sources in Game Mode 1. I tested this extensively by zooming in using the VP50, and then zooming in on my monitor, to get a total of roughly 8x zoom so I could see the "pixels" up close. It looked entirely like it was keeping every scanline at the same height.

I did another test. When a typical VCR displays a menu over a blue screen, such as the menu for programming the record timer, this is 240p output. However, when it displays the same OSD over playback video (such as the channel number), it is 480i. This can be observed on an ordinary CRT by displaying the TV's OSD on top of the VCR image, because the TV's OSD flickers or not based on the video mode of the TV, and the TV's OSD is a good test case because of the high-contrast text.

In the VP50 set to Game Mode 1, I observed vertical "vibration" when the VCR's text displayed over a video, but no vibration at all when the VCR was displaying text over a blue screen.

In fact, I discovered this by accident. When I use a VCR with the VP50 I use Game Mode 1 because the increased softness masks the artifacts inherent in the low-quality source. When Game Mode 1 is used on most sources (480i), there is vertical bobbing because there is no motion-adaptive deinterlacing. However, I noticed it looked completely still on the blue screen with text of a VCR. I hooked up the VCR to a CRT TV and sure enough, it was 240p.

I also compared the output of a PS1 game, played on a PS2 over component, through the VP50 with Game Mode 1, and through the XRGB-2+. The display was, in fact, a bit better with the VP50. (For S-video and composite, the XRGB-2+ was better.)

I'm very confident that the VP50 correctly handles 240p in Game Mode 1, and treats the scanlines in every field as being in the same vertical location. There was frame judder in S-video and composite, which I suspect has more to do with the source quality (NES, SNES, and N64 consoles) than the fact that the source is 240p. Both the Algolith Flea and the PDI Deluxe card also had trouble receiving these sources.

I still think the bobbing effect is just a deinterlacing artifact caused by the fact that the VP50 is using the diagonal interpolation algorithm to fill in the "empty" scanlines, and noise in the signal. The diagonal algorithm decides the angle of interpolation. I'm speculating that it uses "hard thresholds", so when certain pixels are certain brightnesses and/or colors and then one is just very slightly brighter the next field--something that could be caused by noise--the angle is different, and that causes the sparkly effect. If it were bobbing caused by alternating the heights of scanlines, it would be far more pronounced. Which is exactly what is observed when viewing a still image in a 480i signal in Game Mode 1.

You could be right about this, but does it really matter ?

For 240p input the Gamemode 2 looks much more solid (& better) (especially on horizontal 1-pixel lines) than Gamemode 1, but it produces bad deinterlacing artifacts.

All I know and can say for sure is that my previous processor had the stability of the VP50's Gamemode 2 and it didn't produce any deinterlacing errors like the VP50 does in Gamemode 2.

I'm not a 100% sure if I can concur with your "correct 240p interpretation" theory. If you take a single horizontal white line (e.g. the outlines of the power bar in any beat'em up) and pause the game, so there's no movement on screen, why should this line appear shaky if the VP50 does not alternate between "interpolation on line down" and "interpolation one line up" ?

Anyway, it's just a shame that ABT passes out on this one, as it should be so easy to do it right. Keeping the lines on the same level and just linedoubling to the pixel line below all the time without any fancy deinterlacing would to the trick perfectly.

Tobias

A small overview about relevant threads:

HDTV Game Lag VP shootout: XRGB-3 vs iScanVP50:
http://www.avsforum.com/avs-vb/showthread.php?t=747832

Is there any video processor that provides a scanline overlay?:
http://www.avsforum.com/avs-vb/showt...ghlight=gaming

New DVDO iScan VP50PRO:
http://www.avsforum.com/avs-vb/showt...889884&page=83
(read starting around #2467)


they're a compromise, but for sure not the best possible solution to our gaming needs.

Just think about it: the processors can't (or won't) analyze if the incoming signal is 480i or 240p, so a machine like the XRGB2(+) needs a deinterlacing mode which can handle both modes adequately.

I think we both know what a 240p distinguishes from a 480i signal. Both signals have 60 fields, each field 240 lines. On a 480i signal the 240p fields have alternating line offsets (even/odd), while the 240p signal keeps the fields' offsets on the same line. This is why you get those visible dark scanlines on CRT based displays.

To keep text-based 480i games readable the XRGB2 has to alternate the line offsets of the incoming fields. Also the XRGB series doesn't do any motion adaptive deinterlacing, it's a simple video mode - this means every field is processed on it' own, not taking into account what the next or prior field looks like (this is also the only way why the XRGB2 is cheap AND has a very low latency). The odd fields get interpolated (smoothly linedoubled if you like) a line down. The even fields get interpolated a line up.

The very same thing happens on a 240p signal. If you take a 1-line horizontal line (let's say a laser shot in Gradius) it's clearly outlined between two dark scanlines on a CRT display, because the fields are kept at the same line offset. On the XRGB2 or the VP20/30/50s Gamemode 1 you basically get a solid inner part of the laserbeam, but vibrating horizontal edges, especially visible when such details are coming fading in or out. And while a shot is hardly visible for more than a mere second, it gets rather annoying on life bars or the score display.

I absolutely know what the XRGB2 does look good at the first sight (and second), but there's no technical reason why a linedoubled 240p signal can't be as steady, smooth and razorsharp at the same time like the output from an emulator. The VP20/30/50/50pros Gamemode 2 shows this wonderful with non-moving input signals (but of course has the wrong approach for moving pictures from a non-interlaced source).

Thanks for the info on the XRGB2.

You didn't answer my question in regards to DScaler, however. Is there anything wrong with its Old Game mode other than a lack of scanlines and slight smoothing of pixels? I've been quite satisfied with its motion, and I seem to remember comments from the person who programmed it that it properly processes a 240p signal.

It's been quite some time since I used DScaler, though I still have my Sweetspot card installed. From my memory I would say that the "old game" mode in Dscaler is quite similar to what the XRGB2 does, but I wouldn't lean out before giving it another try. Maybe I'll do so the coming days.

I remember that switching between particular BOB videomode and old game didn't make a huge difference.

Anyway, the VP50pro with a designated 240p mode would be easily outperform all other videoprocessors in a heartbeat, especially since DScaler introduces quite some lag. This is way I started the discussion in the 50pro thread and was quite happy when Dale picked up the idea.

Lag? Can't say I've noticed any when using Old Game, and I seem to do okay on fighting games. Have you ever measured it?

No, but from the "feeling" in Danmaku Shoot'em ups considerably slower than the VP50's gamemode 1 which is about the same as the XRGB2+.

Just to chime in, I've never had any particular lag with any of DScaler's deinterlacing modes.

What processor were you using in your tests, Fudoh?

I'm using an X2 4200+ and don't notice anything.

I use an XRGB-2+ with my VP50 (non-Pro). It isn't perfect, but it's the best I know how to do, and until something better comes along, I feel pretty good about it. The image stability thing isn't ever a distraction on my favorite games, and it's, I don't know, spiritually soothing or something to see the pixel art outlined and clarified by scanlines as god intended. The artists at the time may have wished scanlines didn't exist, but I grew up with them, and now they're a distinctive part of the look of those games, and I itch when they aren't there. The slightly better Game Mode 1 of a VP50 Pro isn't enough to pry the upgrade price out of my tight little fists unless I can keep my scanlines, too.

The fact that the shadows are displayed only every other frame is why you see combing on the VP50. The motion sensing in the deinterlacer compares fields which are 2 field periods apart (since they are spatially coincident). When it does this with a signal such as you describe above, it senses no motion and you get combing. The VP50's deinterlacer does also look for differences between fields which are 1 field apart, but its threshold is too high for the type of shadow scenes you're having problems with.

The correct solution for this is to simply treat the input as 240p rather than 480i and not deinterlace it at all. I'm looking into this, and I have it partially working on a VP50Pro, but there are still some issues with a VP50.

I thought you might appreciate the insight into why the combing in occurring on the VP50 on this type of game.

- Dale Adams

Highly appreciated


I know, but since many games run in "240p mapped onto 480i" we need a forced 240p moed instead of an auto detection. I mentioned this in the 50pro thread before though.

Thanks for your help !

Maybe they can just add a "Gamemode 3" which acts like that.

It's on our list to add a 240p input mode to the Lumagen Radiance. Right now 240p gets displayed as 480i. We will be adding a proper 240p input mode to the Radiance for video games. I'm not sure of the schedule for adding this feature.

Best regards,
Randy Freeman

Wow, I never though even one company would seriously look into providing a proper 240p mode, much less 2. Excellent.

Hey everyone,

I was curious as to whether or not I should go with a video processor and/or receiver. I have the Sharp 52SE94U and will only be using it for movie and games. No plans for cable or satellite hookup. I have an Onkyo HD805 HD-DVD player (HDMI), PS3 (HDMI), Xbox360 (Component w/ optical out), Wii (Component), PS2, (Component). I mainly want better video quality for my Wii and PS2. Should I get a receiver that does HDMI upconversion, a video processor for Wii and PS2, or a combination of both? I'm trying to spend as little money as possible.

Thanks in advance,
Plums.

You should define what "as little money as possible" is in numbers...

Well, I would a good scaler with being able to switch between 2 devices. If I'm going to get a a dedicated video processor and a pre/pro receiver I would want to spend a max of $2k. But if I'm just getting one I'm willing to spend around $1,500.

Thanks,
Plums.

A small overview of deinterlacers and scalers available for videogame processing:
http://retrogaming.hazard-city.de/

Wait until EDGE comes out. According to DVDO it will come with Game mode - you can preorder...

The Edge, just like the VP50 and 50pro is not able to process a 240p videogame signal. It's fixable by a FW upgrade though, but it has to be done before the Edge can be recommended to gamers.