Ketsui: Kizuna Jigoku Tachi Extra (XBox 360)

[Animaitor]

Thanks emphatic! I guess the pictures you took are taking with the gear and settings you mentioned on your previous post, right? I'm having a little bit of a problem to get the image to display correctly, even by running the game from the 360 at 640X480, sending the video to a Super Emotia through VGA and from here to the XRBG-3 to my LCD. I also agree with some people about the filter. Near Neighbor doesn't seem to provide me with the best results. If it helps, Fudoh and antares wrote a couple of pages back their findings:

Does anybody know what's the original resolution of the PGM hardware ?? While younger Cave games like Futari or ESP2 are running at 320x240 which makes it easy to map the emulation onto a 720p screen (480x640 with borders all around), Ketsui doesn't offer this option. It also seems to have non-square pixels. With the overscan setting set to 90 or 100% the scaling errors (wavy scrolling) are obvious on both the X- and the Y-Axis. With the overscan set to 83% the waving on the horizontal can be eliminated, while the vertical axis looks good (but not perfect) on this setting. Overall 83% seems to be the best setting though if one doesn't mind the smaller playing area. (Applies to Yoko 720p, Nearest Neighbour only).

Further more, the conversion feels a bit laggy.

I'm playing via scart rgb in 480i and setting the resolution to 93% will eliminate the horizontal waving completely (tate mode). It's still noticeable vertical but I think overall that's the best setting when playing in 480i.

Also I haven't noticed any lag at all when playing with scart rgb cable.

I'll keep testing today and if I get good results, I'll upload some pictures as well.

[Fudoh]

The 83% I found to be best looking is for yoko in 720p only. For any other resolution just check the various settings (just press the back button during play, brings up the video options). It's basically just a mathematical equasion. On a 720p output the 100% area is 540x720 pixels in size. 83% of that are 448x598. This elimates any scaling artefacts on the horizontal axis (since 2x224 = 448). If you want to eyeball the result, take a look at the grid on the bottom of the screen (where the bombs are shown), you'll notice that the vertical grid strips are uniform on a certain scaling percentage, while they vary in width on all other settings.

If playing on a 480p output in Tate 93% (= 448 pixels active image width) should look best, but you get a bit of underscan border around the actual playing area. Bicubic and Bilinear filtering eliminates all the scaling artefacts. Unfortunately 5pb didn't properly implement the N.Neigbhour and Lanczos filtering - this could have looked better...

Without trying myself I would recommend 480i/480p with 93% on a CRT. On an Emotia/XRGB setup I would try 100% along with Bicubic sizing, possibly pumping up the sharpness on the XRGB-3 a bit!

Nevertheless, I'm amazed how good Ketsui actually is. The first three levels are a blast to play and I'm slowly getting used to the 4th level as well

[StarCreator]

If playing on a 480p output in Tate 93% (= 448 pixels active image width) should look best, but you get a bit of underscan border around the actual playing area. Bicubic and Bilinear filtering eliminates all the scaling artefacts. Unfortunately 5pb didn't properly implement the N.Neigbhour and Lanczos filtering - this could have looked better...

I'm not sure how you can implement nearest neighbor "wrong" - when not dealing with an exact multiple of the source resolution (which will always happen in Ketsui since the source doesn't use square pixels) you'll have scaling artifacts simply because it's copying the color of the source pixel with no alteration, just like the algorithm specifies.

As for Lanczos, Ketsui probably uses a small window value so that the filter won't take more than x processing time, since this has to be done in realtime and no one but 5pb really knows how much CPU muscle the game itself needs. A larger window would have made it look better, but it also might have been too much for the 360's CPU to handle while still rendering at a full 60fps.

[SP3RRFEUER]

but it also might have been too much for the 360's CPU to handle while still rendering at a full 60fps.

Think it over

[yosai]

running the game in 24kHz with the UVC on my NANAO MS9 looks... wait for it.. as good as, if not better than the PCB.

Mate if the PCB looks that blurred on your cab something is not working quite right.

[cools]

running the game in 24kHz with the UVC on my NANAO MS9 looks... wait for it.. as good as, if not better than the PCB.

Mate if the PCB looks that blurred on your cab something is not working quite right.

Running a 15k game stretched up (doubled) to 31k then downscaled to 24k... Cmon emphatic, that's like applying an AdvanceMAME filter to it!

EDIT: http://www.pici.se/588331/ <- that's terrible!

[emphatic]

cools, yosai: My photo skills really are subpar, so the blur is my camera+jpeg really. I totally get the AdvanceMAME comparison though, I'd rather have a raw unfiltered picture.

[Fudoh]

I'm not sure how you can implement nearest neighbor "wrong" - when not dealing with an exact multiple of the source resolution

this is a bit hard to explain, but still very obvious if you compare Ketsui's NN implementation with the bicubic/bilinear filters. Take a look at the vertical axis of the game. Here you end up with a scaling factor between 1,33 and 1,5 depending on the zoom level (this is true for all filtering modes). Ketsui applies fixed scaling factors to every line of the visible area (e.g. line 100 of the screen remains always unscaled with 1,0). This causes the graphics to "wobble" while scrolling, because the scaling factors for a certain line of the background are changing all the time while this one line is panning down the screen (the graphics are panning behind the scaling matrix). If you implement NN correctly you apply a fix scaling factor to each line (of the graphics, not of the screen), but you have to make sure that the scaling factor stays the same for this particular line of pixels for the whole time this line is moving across the screen. In Bicubic/bilinear filtering they did this. That's why there's absolutely no wobble in those modes.

Sorry if this sounds a bit confusing, but take MAME where this is done correctly. You can also take Raiden Fighters Aces. In the first japanese (unpatched) version, they did it wrong, but it was fixed with one of the updates.

[StarCreator]

this is a bit hard to explain, but still very obvious if you compare Ketsui's NN implementation with the bicubic/bilinear filters. Take a look at the vertical axis of the game. Here you end up with a scaling factor between 1,33 and 1,5 depending on the zoom level (this is true for all filtering modes). Ketsui applies fixed scaling factors to every line of the visible area (e.g. line 100 of the screen remains always unscaled with 1,0). This causes the graphics to "wobble" while scrolling, because the scaling factors for a certain line of the background are changing all the time while this one line is panning down the screen (the graphics are panning behind the scaling matrix). If you implement NN correctly you apply a fix scaling factor to each line (of the graphics, not of the screen), but you have to make sure that the scaling factor stays the same for this particular line of pixels for the whole time this line is moving across the screen. In Bicubic/bilinear filtering they did this. That's why there's absolutely no wobble in those modes.

Sorry if this sounds a bit confusing, but take MAME where this is done correctly. You can also take Raiden Fighters Aces. In the first japanese (unpatched) version, they did it wrong, but it was fixed with one of the updates.

So basically, you're saying, they should have scaled the game's assets individually, then render the game screen with the resized assets, rather than render the game screen at the original resolution then perform a single scale operation (which is what appears to be the case currently)?

[Fudoh]

I'm not saying how they could have done it better, just that they could have done it better Prescaling all the graphics would only work for a single resolution with a single zoom setting. I was simply suggesting the lock the scaling matrix to the background. This would cure the visible wobble in the background. The sprites would still look the same, but I think it's the silent moments in which you look around which are a bit distracting. I wanted to do some Ketsui video HD captures anyway, so maybe I'll do a scaling comparison between various games running on the same resolution incl. MAME....

[Animaitor]

I wanted to do some Ketsui video HD captures anyway, so maybe I'll do a scaling comparison between various games running on the same resolution incl. MAME....

Yes please. That would be lovely!

[Shelcoof]

I wanted to do some Ketsui video HD captures anyway, so maybe I'll do a scaling comparison between various games running on the same resolution incl. MAME....

Yes please. That would be lovely!

+1

[roker]

I tried to hold back, but I can't ...

why does it seem as if every topic here sooner or later delves into a 15 page discussion on resolution?

[MX7]

I tried to hold back, but I can't ...

why does it seem as if every topic here sooner or later delves into a 15 page discussion on resolution?

I know. I guess everyone who isn't masturbating over scanlines (or having an angry wank over the lack thereof) is actually playing the game. Which is fair enough.

I haven't had much time to play the game yet, but I have enjoyed immensely what I've played. The game is a lot harder than I remember. After spending so much time on Death Label, the bosses seem pretty easy compared to the stages. I really need a bigger screen to make serious headway, I think. Futari 1.5 Original has softened me up

[StarCreator]

I was simply suggesting the lock the scaling matrix to the background. This would cure the visible wobble in the background. The sprites would still look the same, but I think it's the silent moments in which you look around which are a bit distracting.

I don't think the sprites would look the same if the scaling worked as you proposed - rather, I would think the scaling errors would move through the sprites like it were tides in the ocean, which would be arguably more noticable than the problem this method would be trying to fix. Probably the best way to go, if not prescaling each individual sprite, would be to scale each graphics layer separately and composite them post-scaling... though it makes me wonder if you'd lose a little graphical accuracy compared to the PCB doing it that way.

[MachineAres 1CC]

For 1 you have to clear both the standard loop and the URA loop, more details here: http://shmups.system11.org/viewtopic.php?f=5&t=5123

How does this make sense? Elaborate more? I've beat both standard and URA loops with both ships, and DOOM never appears. I tried doing normal loop and then getting the URA requirements on the loop itself, and just got a game over, it didn't carry into another loop. What exactly does "clear both loops" mean?

[nZero]

For 1 you have to clear both the standard loop and the URA loop, more details here: http://shmups.system11.org/viewtopic.php?f=5&t=5123

How does this make sense? Elaborate more? I've beat both standard and URA loops with both ships, and DOOM never appears. I tried doing normal loop and then getting the URA requirements on the loop itself, and just got a game over, it didn't carry into another loop. What exactly does "clear both loops" mean?

Which part is confusing? If you make it to the end of the Ura loop in Arcade mode, Doom shows up after you destroy the Stage 5 boss. If you want to do it the hard way, with the EX Options on defaults, clear the first loop without dying or bombing and you'll be taken to the first stage of the Ura loop after beating stage 5. If you just want the achievement, go to Config EX and set Loop to Ura and Continue At Doom to On, and then start a game in arcade mode and credit feed through 10 stages.

[Chacranajxy]

Got the game yesterday, and I've played for about an hour and a half. I can nearly kill off the third stage's boss, but not quite... game is pretty damn tough, but ridiculously satisfying.

[MachineAres 1CC]

For 1 you have to clear both the standard loop and the URA loop, more details here: http://shmups.system11.org/viewtopic.php?f=5&t=5123

How does this make sense? Elaborate more? I've beat both standard and URA loops with both ships, and DOOM never appears. I tried doing normal loop and then getting the URA requirements on the loop itself, and just got a game over, it didn't carry into another loop. What exactly does "clear both loops" mean?

Which part is confusing? If you make it to the end of the Ura loop in Arcade mode, Doom shows up after you destroy the Stage 5 boss. If you want to do it the hard way, with the EX Options on defaults, clear the first loop without dying or bombing and you'll be taken to the first stage of the Ura loop after beating stage 5. If you just want the achievement, go to Config EX and set Loop to Ura and Continue At Doom to On, and then start a game in arcade mode and credit feed through 10 stages.

That did work, setting the loop setting to "URA," but it clarifies my issue. You can't get the achievement for beating DOOM with No Bullet on.

[adversity1]

You can't get the achievement for beating DOOM with No Bullet on.

lolol

try turning the douche setting off too thx

[EOJ]

time will tell if it continues to decline now that the port is out.

Just following up on this point, but it looks like post-port Ketsui PCB prices may be stronger than I expected, one auction is up to 70K yen and counting (bare board w/ copied art), ending soon:

http://page6.auctions.yahoo.co.jp/jp/auction/f91237409

Perhaps the previous 51K yen one was just an anomaly.

[yosai]

time will tell if it continues to decline now that the port is out.

Just following up on this point, but it looks like post-port Ketsui PCB prices may be stronger than I expected, one auction is up to 70K yen and counting (bare board w/ copied art), ending soon:

http://page6.auctions.yahoo.co.jp/jp/auction/f91237409

Perhaps the previous 51K yen one was just an anomaly.

Having an xbox360 inside a cab is second best to having a PCB in there.

As awareness of the game increases the demand for the PCB should also increase. Or some shit.

[EOJ]

As awareness of the game increases the demand for the PCB should also increase.

This hasn't happened with other recent Cave games that have been ported (Deathsmiles, Futari 1.5), so I doubt that.

[JAPJAC]

I doubt one single person's recent blog opinion on the worth of the PCB will do a 'Brandon Sheffield' but 8 years later and in reverse.

To say that they would have to think about buying a Ketsui PCB for 65 quid today, slightly more than a poster, is just laughable and/or bizarre.

Ketsui PCB will and should retain it's value as it's arcade machine/RGB low res. ready for starters, regardless of a quality home conversion. A ton was taking the mick but 70-80 grand is still fair imo. 350 quid was a steal and 65 quid is never going to happen.

Cheers.

[Elixir]

As awareness of the game increases the demand for the PCB should also increase. Or some shit.

This is what PCB collectors really think.

[bcass]

Having an xbox360 inside a cab is second best to having a PCB in there.

It is? Correct me if I'm mistaken, but the PCB doesn't have X-Mode, online Leaderboards, uploadable/downloadable replays, training modes, additional options or optional remixed soundtrack.

[charlie chong]

EVEN AFTER THE PORT HAS COME OUT YOU GUYS CAN'T STOP BITCHING ABOUT PCB OWNERS
JUST BE HAPPY WID YOR EGGS BOGS PORT YOU TURDS
ALSO BCASS WHEN YOU GONNA LEARN KNOW ONE CARES ABOUT WHAT YOU THINK COS YOU A POINTLESS IDIOT MORE CONCERNED WITH SUPEFLUOS DETAILS THAN ACTUALLY PLAYING THE GAME.

CAN WE TALK ABOUT KETSUI'S EXCELLENT BULLET PATTERNS THAT GO SIDEWAYS AND ARE MORE FUN TO DODGE THAN BORING PATTERNS OF MUSHIMESAMA GAMES RATHER THAN DISPLAYS AND JEALOUSY PLEASE??

[Skykid]

Lol, Charlie you crack me up man.

[Fudoh]

KETSUI'S EXCELLENT BULLET PATTERNS

Ketsui's bullet patterns are a work of art indeed. Those waves with changing bullet speeds within the pattern - fabulous. Dying with a smile on your face

Who designed the bullet patterns for Ketsui ? I mean, there a few moments in Mushihimesama and others in which I believe to recognize the design, but overall I'm amazed at how great the Ketsui patterns are and that the succeeding Cave shoot'em ups didn't built on that.

[charlie chong]

i feel the espgaluda 1 + 2 patterns carried on the tradition somewhat in places .. it was actually down to a programmer because ikeda apparently always wanted danmakus that were straight but the coder apparently had enough and wanted to try some different style patterns.

so how do you react to them ?? do you find them hard to dodge?
personally i think they make more sense for me to dodge thriough than any game cave has released since!

heres the quote from an adverse translation on cave stg
・Ketsui - Kizuna Jigokutachi "Uniform speed and a direct line" were the ironclad rules for Dodonpachi's danmakus, but Ichimura, the programmer for Daioujou, eventually whined about this so...(lol). This was a title where we really pushed his taste to the forefront and did "whatever we wanted".