FPGA Based Videogame System

[Keatah]

They should.

[Andromeda Stardust]

I am curious if the NT Mini custom 8bitdo firmware maps the 8bitdo controllers differently to how the original new and old 8bitdo firmwares operate. Specifically, I want to know which button mapping the Analogue custom firmware uses and if it allows turbo function or other usage of the Y/A/L/R buttons. I may have to test it myself...

Okay I figured it out. The custom v1.23 firmware by Analogue uses Y/B and completely disables the turbo function on X/A. I guess the Analogue guys are not fans of turbo controls. Reverting back to v1.10 beta4...

[atmn]

Kevtris, noticed that you said in the live stream that the SFC data pad worked with the Intellivision and Colecovision cores.

Does this work aswell? This is the Famicom Network version.

 

Edited March 1, 2017 by atmn

[kevtris]

I noticed the save thing as well, there's a Super Mario Bros 3 save hack that doesn't work since it expects SRAM to be 0'd out. Neat to hear that that is being worked around.

 

The weird thing is that submapper is set on my StarTropics rom as instructed (I used Nintendulator to do it), and music works, but if I save the game, go to menu to dump the SRAM to a file, then power off + power on the console and load up the game again, the save doesn't work. But if the iNES 2.0 header isn't present, it saves yet won't play music.

 

Sorry if I'm just repeating what you already know! Just want to make sure

yeah I am pretty sure I fixed it. I did the same test after I fixed it tonight and it seems to work. For completeness, I played ST1 then ST2 and ST1 again, back and forth and it seems solid now.

[kevtris]

Kevtris, noticed that you said in the live stream that the SFC data pad worked with the Intellivision and Colecovision cores.

Does this work aswell? This is the Famicom Network version.

 

I have one of these but it doesn't work yet. I could make it work on the cores though. I will add it to the todo.

[atmn]

I have one of these but it doesn't work yet. I could make it work on the cores though. I will add it to the todo.

 

Lovely, my FC Network pad would be perfect for those systems! Thanks!

Edited March 1, 2017 by atmn

[Andromeda Stardust]

Kevtris, noticed that you said in the live stream that the SFC data pad worked with the Intellivision and Colecovision cores.

Does this work aswell? This is the Famicom Network version.

 

 

 

I have one of these but it doesn't work yet. I could make it work on the cores though. I will add it to the todo.

I'm curious if a simple expansion port to NES adapter would work with these. Or is the bit order of the buttons different on these???

[kevtris]

 

 

I'm curious if a simple expansion port to NES adapter would work with these. Or is the bit order of the buttons different on these???

sure it would work. the first 8 buttons are in the usual order, this is because you can plug this into the expansion port and use it as another controller. Famicom games generally support this, but NES only games don't. I have not checked out the bit order of the other buttons but it isn't too difficult to do so. I have a program specifically for this purpose that will work.

[Andromeda Stardust]

sure it would work. the first 8 buttons are in the usual order, this is because you can plug this into the expansion port and use it as another controller. Famicom games generally support this, but NES only games don't. I have not checked out the bit order of the other buttons but it isn't too difficult to do so. I have a program specifically for this purpose that will work.

Reason I asked is because the SFC controllers have the extra 8 bits though four of them are key bits and don't register presses. So I imagine the SCF data controllers would have the bits pushed back some on the special buttons compared to the FC models.

 

I have a red/gold Famicom Joycard I with the honeybee logo on it. Cable is short but I've got some controller extensions from Tototek. I bought my FC Joycard on eBay last year after I fell in love with the half-working graphite gray NES Joycard II I bought used for cheap, and later refurbed the turbo contacts with sandpaper and dielectric grease. Oxidation was so heavy and the gold plating was mostly gone so good old 600 grit to expose the copper and grease to seal it off. The NES Joycard have audio passthrough and a headphone jack of all things, so kids can play in the dark at night without waking their parents, LOL.

 

Back to the Famicom expansion controllers, surprisingly many NES games from Japanese studios support them though western developed games generally don't. Super Mario USA is one of few Famicom carts I own that it doesn't work with, possibly due to it's being a port of an originally US exclusive game. Still odd Nintendo didn't add it in. Many games read D1 in addition to D0 but it's really hit or miss.

[Keatah]

A cool thing coming from FPGA is that people are finally beginning to accept 1080p and LCD as the new gaming standard. The complaints about LCD not replicating the old analog CRT "effects" and special artifacting tricks are beginning to die off. The muss and fuss of connecting analog consoles to digital sets is subsiding. FINALLY!! Matrix addressed displays are superior all-around.

Edited March 1, 2017 by Keatah

[Fafner]

Just realized what I'm gonna do with my Zimba 3000 when it gets here!

I have a Mame cabinet that was made from a Neo-Geo cabinet.

Zimba 3000 with a Neo-Geo set in the cabinet is gonna be glorious!

 

Kevtris!

I have a question about the controller adapters for the Analogue NT.

I've heard you mention using the NES cart port for it but as many have noted the adapter is a bit wobbly.

Is there potential for that to be an issue?

[Andromeda Stardust]

A cool thing coming from FPGA is that people are finally beginning to accept 1080p and LCD as the new gaming standard. The complaints about LCD not replicating the old analog CRT "effects" and special artifacting tricks are beginning to die off. The muss and fuss of connecting analog consoles to digital sets is subsiding. FINALLY!! Matrix addressed displays are superior all-around.

I don't think anything is "dying off" just a new way to play, that's all. I love my AVS as an HD alternative to NES/FC. Retron5 / Retrofreak (see also: YAEB) not so much. Also 720p is every bit as good as 1080.

[rezb1t]

Also 720p is every bit as good as 1080.

If your monitor/TV's native resolution is 1080p then the signal being 720p results in scaling, which leads to more latency. Most TVs and monitors *should* do a good job of this but it was still enough to be noticeable across 3 displays in my case.

 

720p on 4k displays should be great though. The AVS is decent at its price point but I ended up feeling disappointed with the lack of updates, and every new beta firmware just seemed to break something else. No harm done though, their value holds really well, I ended up selling mine for a small profit. I hope the situation changes for the better in the near future.

[Wolf_]

If your monitor/TV's native resolution is 1080p then the signal being 720p results in scaling, which leads to more latency. Most TVs and monitors *should* do a good job of this but it was still enough to be noticeable across 3 displays in my case.

 

720p on 4k displays should be great though. The AVS is decent at its price point but I ended up feeling disappointed with the lack of updates, and every new beta firmware just seemed to break something else. No harm done though, their value holds really well, I ended up selling mine for a small profit. I hope the situation changes for the better in the near future.

If 4k isn't supported via fpga then 720p is ideal as it is an integer scale and avoids distortions, although the tv or an upscaler still needs to process the image which almost certainly will add latency unless you use something like a 4k capable OSSC.

 

That being said obviously the highest resolution your display can handle that is an integer scale of the original resolution is always ideal. As someone that invested in 4k for pc gaming I sure would appreciate not having to keep a 1080p display around while I wait for fpga 0 latency native rendering tech to catch up with modern displays.

[Andromeda Stardust]

If 4k isn't supported via fpga then 720p is ideal as it is an integer scale and avoids distortions, although the tv or an upscaler still needs to process the image which almost certainly will add latency unless you use something like a 4k capable OSSC.

 

That being said obviously the highest resolution your display can handle that is an integer scale of the original resolution is always ideal. As someone that invested in 4k for pc gaming I sure would appreciate not having to keep a 1080p display around while I wait for fpga 0 latency native rendering tech to catch up with modern displays.

I think you'll be waiting a long time for 4k native output. By then we should have CRT solder masks and bloom effects down to a tee. Until then, 720p "Trumps" 1080p based on integer scaling, references to 45th president notwithstanding.

[cacophony]

Why is 720p output better than 1080p output for 4k displays?

 

720p * 3 = 4k

But...

1080p *2 = 4k

 

They are both integer scales to achieve 4k, but in the 1080p case you get the more accurate pixel aspect with your starting integer scale (either 5x height by 6x width or 4x heigh by 5x width). Plus it might introduce less lag to do the 2x scale compared to the 3x.

Edited March 2, 2017 by cacophony

[phoenixdownita]

Why is 720p output better than 1080p output for 4k displays?

 

720p * 3 = 4k

But...

1080p *2 = 4k

 

They are both integer scales to achieve 4k, but in the 1080p case you get the more accurate pixel aspect with your starting integer scale (5x height by 6x width).

240 * 3 = 720

240 * 4.5 = 1080

 

the issue is the source material, most if not all 8bit/16bit and 32bits out in 240p (exceptions here and there, some games have 480i sometimes, some console have weird 288 as well) , you do the math for the source material as it stands today (well 20+Y ago really but I digress).

 

4K direct works for both because

720 * 3 = 2160 -> 240 * 9

1080 * 2 = 2160 -> 240 * 9

 

(4K being usually 3840 pixels × 2160 at 60p in these discussions)

 

but if you go first at 1080p then you have the 4.5 non integer factor to deal with first, hence better have had a 720p to feed to the 4K "internal scaler", it is really just a line tripler in the 720p case and a line doubler for 1080p .... both of which can be accomplished with just one line of buffer or if you prefer one line [of the original signal] of lag or 70 usec (0.069 msec = 16.66/240).

[cacophony]

Not sure I follow that logic. Many of us prefer 1080p output for PAR reasons and the larger image, and doing an integer scale of that exact output to 4k would have exactly the same properties.

 

I don't want the 240 pixels to perfectly fill the display because some of it was intended for overscan. I'd much rather play with a larger image that has 5% of the top and bottom cropped. It also yields a 1.2 pixel aspect, which is much preferable to the 1.0 or 1.33 you get with 720p integer output.

Edited March 2, 2017 by cacophony

[rezb1t]

I think you'll be waiting a long time for 4k native output. By then we should have CRT solder masks and bloom effects down to a tee. Until then, 720p "Trumps" 1080p based on integer scaling, references to 45th president notwithstanding.

I strongly disagree. 5x vertical/6x horizontal is integer scale *and* looks much better than either the too skinny 3x or too fat 4x you need to settle with on 720p

[cacophony]

I strongly disagree. 5x vertical/6x horizontal is integer scale *and* looks much better than either the too skinny 3x or too fat 4x you need to settle with on 720p

 

Exactly this. My largest complaint with the AVS is that I needed to play in a non-integer scale (3rd tick) to have the image look natural to me. For me, the biggest benefit of the Nt Mini prior to the jailbreak was the 5x by 6x mode.

[phoenixdownita]

Not sure I follow that logic. Many of us prefer 1080p output for PAR reasons and the larger image, and doing an integer scale of that exact output to 4k would have exactly the same properties.

 

I don't want the 240 pixels to perfectly fill the display because some of it was intended for overscan. I'd much rather play with a larger image that has 5% of the top and bottom cropped. It also yields a 1.2 pixel aspect, which is much preferable to the 1.33 you get with 720p output.

To each his own, if you are OK "wasting" 15% of your screen real estate it's fine by me and it should definitely be an option.

The overscan area btw is usually beyond those 240p. NTSC is really 525 lines (interlaced) or ~262.5p.

 

The actual overscan of broadcasting (and to compensate for crappy circuitry in 1930 TVs) is much more (5% to 10%) no questions there.

[cacophony]

To each his own, if you are OK "wasting" 15% of your screen real estate it's fine by me and it should definitely be an option.

The overscan area btw is usually beyond those 240p. NTSC is really 525 lines (interlaced) or ~262.5p.

 

The actual overscan of broadcasting (and to compensate for crappy circuitry in 1930 TVs) is much more (5% to 10%) no questions there.

 

From the comparisons I've done the amount of image lost is actually a tiny bit less than on most CRTs. Lon Seidman has a good comparison using RC Pro Am here:

https://youtu.be/_y5XJugZRks?t=8m5s

 

I would consider the default black bars on the top/bottom of AVS output as "wasting" screen real estate because I'd prefer to have the image fill the screen as much as possible while keeping things accurate to the original experience.

Edited March 2, 2017 by cacophony

[phoenixdownita]

 

From the comparisons I've done the amount of image lost is actually a tiny bit less than on most CRTs. Lon Seidman has a good comparison using RC Pro Am here:

https://youtu.be/_y5XJugZRks?t=8m5s

It's a question of habit and how you used to play way back when.

I remember playing my VCS on a crappy old set and whatever it gave me I was fine, but then in PS1 time I had a 1084 Monitor via Scart and as I played both 50Hz and 60Hz I found a reasonable compromise with the V-stretch control so I would not need to change it every time. The H-stretch also was set to suit my taste.

 

With that setting I believe I may actually have thrown down the drain whatever pixel-ratio or other accidental disaster the various companies threw at me as I "fixed them" to suit my taste anyway (PS1, Saturn and N64 all got the same treatment, big N being the assholegenius of signal generation with their weird pixel aspect ratios all the way to the GC, wasn't that a 20:23 or something for the GC).

 

I still have issues to this day playing my SNES vs say my MD because the aspect looks "different" and somewhat not to my taste, I cut a lot of slack to the 8bits as they are what they are and honestly it's a miracle they even work at all.

[Andromeda Stardust]

I strongly disagree. 5x vertical/6x horizontal is integer scale *and* looks much better than either the too skinny 3x or too fat 4x you need to settle with on 720p

Except with 5x scaling, you're clipping off 12 scanlines from the top and bottom of 240p signal. I'll admit 4x3 looks a bit wide and 3x3 looks a tad bit skinny, and the 3.5x3 setting is not a bad compromise in 720p on the AVS, even considering chessboard background patterns, which a lot of NES games use to dither colors. Just avoid settings like 3.75 and you'll be okay. The 3/4 and 1/4 skinny wide pixel settings look like ass imo And scanlines are sort of the perfect thickness with 2/3 setting on 720p. Really it depends on the game and display if I use 3.5 or 4 pxel aspect.

 

Maybe I'm trying to justify spending less $$$ on an "inferior" solution such as AVS, since the NT Mini also supports 720p (like AVS) and 480p, but I digress. 1080p does not support integer scaling without padding or cropping the picture. On any HDTV display with built in overscan, 5x4 would be preferable to 6x5, but of cource 1080p desktop monitors don't have overscan baked in.

[Andromeda Stardust]

 

From the comparisons I've done the amount of image lost is actually a tiny bit less than on most CRTs. Lon Seidman has a good comparison using RC Pro Am here:

https://youtu.be/_y5XJugZRks?t=8m5s

 

I would consider the default black bars on the top/bottom of AVS output as "wasting" screen real estate because I'd prefer to have the image fill the screen as much as possible while keeping things accurate to the original experience.

You can minimize the black bars but with current firmware, you cannot remove all overscan region. The latest AVS firmware (v1.2 b8 I believe) still restricts overscan blanking to minimum two pixels top and bottom, maximum 16 pixels top and bottom. Default = 8. I would prefer all 240 scanlines visible but the currentfirm won't allow it. For latest AVS firmware, goto

http://retrousb.com/downloads/

Sort by date and the most recent file is at the top or bottom, I forget which.

 

BTW, you said three ticks. Are you using zero based indexing (leftmost tick = 0) or one based indexing (leftmost tick = 1)?

Using zero based indexing, 0 ticks = 3x3, 4 ticks = 4x3, and 8 ticks (rightmost) = 5x3. The 5x3 setting is useful if you're using an old 4x3 display or can manually set your TV to display 720p in 4x3 aspect. Otherwise, I prefer to use 2ticks right of full left (zero base indexing, 3 ticks if using one based) which is 3.5x3. Any of the odd ticks are just awful. Default is center (four ticks, zero base) at 4x3. I generally tend to use 4x3 on my ASUS monitor and 3.5x3 on my old Sanyo 720p HDTV. The odd ticks look bad and have alias and scrolling issues.

 

EDIT: sorry for notmerging posts; it's a PITA on mobile and if I copy/paste, I run the risk of losing the post. Also I am slightly intoxicated right now, but not enough to forbid coherant posts.