Composite Adapter for MiSTer

For topics which do not fit in other specific forums.
User avatar
legacypixels
Posts: 91
Joined: Sun May 24, 2020 9:20 pm
Has thanked: 25 times
Been thanked: 25 times

Re: Composite Adapter for MiSTer

Unread post by legacypixels »

akeley wrote: Mon Sep 21, 2020 4:16 pm
akeley wrote: Thu Sep 17, 2020 7:40 am If I buy this board, will I see any EXTRA artifacts caused by MiSTer cores + board itself?
I've asked this really simple question some time ago, and nobody has answered it yet. "Yes" or "no" only, please.
It is possible (and likely) you will see additional dot-crawl versus the original hardware. This may or may not be obvious or distracting. This is no fault of the adapter designers, as the MiSTer cores do not provide a key signal need to synchronize the core and adapter clocks. There's an old saying: The man with two watches never really knows what time it is....
US based seller of MiSTer addon boards. Visit https://www.legacypixels.com/mister
dshadoff
Core Developer
Posts: 534
Joined: Sun May 24, 2020 9:30 pm
Has thanked: 19 times
Been thanked: 141 times

Re: Composite Adapter for MiSTer

Unread post by dshadoff »

The challenge with phrasing a question in absolute terms and requesting a binary response is that the question actually needs to be phrased with parameters defining the limits of measurement...

"Is there water on the moon ?" -> yes... no... sort of... depends...

"Is there lag on MiSTer USB with high speed polling ?" -> yes... no... not "per se"...

The answer, relatively speaking (i.e. as far as 99% of the population is concerned) is one thing, but if you want to talk about measurable minutae, the absolute answer is something else.... which can only be discussed in context. If the context itself is confusing or outside of your field of expertise, then probably you are looking for the 99% answer (or you may need to learn more in order to understand why the context is part of the response).
akeley
Top Contributor
Posts: 1303
Joined: Mon May 25, 2020 7:54 pm
Has thanked: 416 times
Been thanked: 399 times

Re: Composite Adapter for MiSTer

Unread post by akeley »

dshadoff wrote: Mon Sep 21, 2020 4:42 pm The challenge with phrasing a question in absolute terms [...]
There is no challenge whatsoever. There either are extra artifacts or there aren't. Your reply is symptomatic of a ridiculous amount of obfuscating noise which was already presented here in response to some basic queries about a product. The only new thing is the added condescension.
legacypixels wrote: Mon Sep 21, 2020 4:27 pm It is possible (and likely) you will see additional dot-crawl versus the original hardware.
Thank you, we are getting somewhere finally. I find it quite unbelieveable it was so incredibly hard though.

The whole issue by the way, is not about the flaw itself but about stubborn refusal to admit that it exists, plus the misleading and dishonest info on the product's webpage.
This adapter allows the connection of your system to a CRT TV or a modern TV with analogue input. This allows the FPGA to provide the same video signal as the original system. The FPGA will skip all scan doublers and video scalers and produce just the pure video signal. That output will then be processed into a composite/S-Video signal using the best-in-class analogue video converter, the AD723 from Analog Devices.
Highlights are mine, but I think it's easy to see how a prospective buyer might get an impression he/she is purchasing a converter which will produce an image which is 1:1 with original hardware . Such buyer has right to know about any possible imperfections, so they might make an informed decision. At least that's how it works in the world outside of this thread.

As it is I will most likely buy one anyway, at least for S-Video, just wish it was possible without all this silliness.
antoniovillena
Posts: 137
Joined: Sun May 24, 2020 8:11 pm
Has thanked: 2 times
Been thanked: 48 times

Re: Composite Adapter for MiSTer

Unread post by antoniovillena »

akeley wrote: Mon Sep 21, 2020 5:37 pm The whole issue by the way, is not about the flaw itself but about stubborn refusal to admit that it exists, plus the misleading and dishonest info on the product's webpage.
I don't know what you win spreading bullshits here and in other forums against me:

https://spectrumcomputing.co.uk/forums/ ... 712#p46712

There is no dishonest info on my product page
akeley
Top Contributor
Posts: 1303
Joined: Mon May 25, 2020 7:54 pm
Has thanked: 416 times
Been thanked: 399 times

Re: Composite Adapter for MiSTer

Unread post by akeley »

antoniovillena wrote: Mon Sep 21, 2020 7:55 pm I don't know what you win spreading bullshits here and in other forums against me:

https://spectrumcomputing.co.uk/forums/ ... 712#p46712

There is no dishonest info on my product page
Coincidence. Interested parties might read my reply there and decide if it's "spreading bullshit" or simply talking about some facts.

As for this forum, your page talks about a flawless product. It's been established it's not a flawless product. End of story, and my participation in this thread.
User avatar
Newsdee
Top Contributor
Posts: 830
Joined: Mon May 25, 2020 1:07 am
Has thanked: 98 times
Been thanked: 209 times

Re: Composite Adapter for MiSTer

Unread post by Newsdee »

As far as I know there is no adapter that will produce perfect composite with MiSTer.

You can't fault antonio for trying to support multiple FPGA boards, propose a solution to MiSTer that works, and getting turned down. Different people will have different preferences. Maybe once more adapters get made there will be an accepted common solution in the framework.
antoniovillena
Posts: 137
Joined: Sun May 24, 2020 8:11 pm
Has thanked: 2 times
Been thanked: 48 times

Re: Composite Adapter for MiSTer

Unread post by antoniovillena »

akeley wrote: Tue Sep 22, 2020 8:19 amEnd of story, and my participation in this thread.
Let's see if it's true
d909
Posts: 35
Joined: Sat Sep 12, 2020 3:21 am
Has thanked: 1 time
Been thanked: 10 times

Re: Composite Adapter for MiSTer

Unread post by d909 »

Antonio's adapter is pretty good. It's actually better than my real NES's composite output, though as described in this thread, there IS a small amount of moving dot crawl. The artifacts on a real NES sit still but are much more visible.

Real NES:
real nes.jpeg
real nes.jpeg (4.25 MiB) Viewed 9801 times
-----
Antonio's adapter:
antonio adapter.jpeg
antonio adapter.jpeg (4.13 MiB) Viewed 9801 times
There is moving dot crawl in the above image, but it's really subtle, particularly when the screen isn't scrolling. I should point out that it's also great in the SNES core but the moving dot crawl is quite a bit worse in the Genesis / Megadrive core. Rainbow artifacts all over the place. That said, Antonio's adapter is still better at this than the others I've tried.
-----

This adapter: https://www.converters.tv/rgb_to_pal/RG ... r/459.html
generic adapter.jpeg
generic adapter.jpeg (4.12 MiB) Viewed 9801 times
The moving dot crawl in this one is a fair bit more visible. All around worse.
-----
I've also tried a couple other adapters. The one js-technology.com sells is pretty high quality, but completely screwed up the color palette when used with the MiSTer for some reason I can't figure out. I also tried a V2 Super Gun from timeharvest on aliexpress, which can also do this conversion, but it's quality is also noticably worse than Antonio's.
ash2fpga
Posts: 237
Joined: Tue May 26, 2020 6:20 pm
Has thanked: 62 times
Been thanked: 28 times

Re: Composite Adapter for MiSTer

Unread post by ash2fpga »

d909 wrote: Fri Oct 09, 2020 8:40 pm Antonio's adapter is pretty good. It's actually better than my real NES's composite output, though as described in this thread, there IS a small amount of moving dot crawl. The artifacts on a real NES sit still but are much more visible.
Thanks for the screenshots! Just curious, have you noticed any differences in some of the textures that game devs [may have] taken advantage of the composite video effects. For example: https://twitter.com/ECUProfessor/status ... 0738290688 (trees in DuckTales)
d909
Posts: 35
Joined: Sat Sep 12, 2020 3:21 am
Has thanked: 1 time
Been thanked: 10 times

Re: Composite Adapter for MiSTer

Unread post by d909 »

ash2fpga wrote: Fri Oct 09, 2020 9:44 pm
d909 wrote: Fri Oct 09, 2020 8:40 pm Antonio's adapter is pretty good. It's actually better than my real NES's composite output, though as described in this thread, there IS a small amount of moving dot crawl. The artifacts on a real NES sit still but are much more visible.
Thanks for the screenshots! Just curious, have you noticed any differences in some of the textures that game devs [may have] taken advantage of the composite video effects. For example: https://twitter.com/ECUProfessor/status ... 0738290688 (trees in DuckTales)
I suppose the best way to answer this question is with screenshots. Antonio's adapter using composite vs RGB. All of these are on a BVM-14F5U. Composite is using a BKM-24N.
duck tales composite.jpeg
duck tales composite.jpeg (3.36 MiB) Viewed 9793 times
duck tales rgb.jpeg
duck tales rgb.jpeg (3.87 MiB) Viewed 9793 times
mega man VI composite.jpeg
mega man VI composite.jpeg (3.67 MiB) Viewed 9793 times
mega man VI rgb.jpeg
mega man VI rgb.jpeg (4.36 MiB) Viewed 9793 times
And here's that mega man VI screenshot using a lower resolution monitor (commodore 1084s-d2) with the same composite:
mega man VI composite Commodore 1084d2.jpg
mega man VI composite Commodore 1084d2.jpg (3.8 MiB) Viewed 9793 times
I tend to think that this last screenshot is a great example for why someone might want to use composite when RGB is a thing.
dshadoff
Core Developer
Posts: 534
Joined: Sun May 24, 2020 9:30 pm
Has thanked: 19 times
Been thanked: 141 times

Re: Composite Adapter for MiSTer

Unread post by dshadoff »

Those are very interesting. Especially the different shades of greens on Duck Tales (not as NTSC artifacts, but as areas of flat colour).
Reminds me of what we saw as differences between RGB and YUV output from the 6260 on PC Engine over the past several months (as the new palette was developed).

Does the composite adapter use RGB as input or YUV ?
d909
Posts: 35
Joined: Sat Sep 12, 2020 3:21 am
Has thanked: 1 time
Been thanked: 10 times

Re: Composite Adapter for MiSTer

Unread post by d909 »

dshadoff wrote: Fri Oct 09, 2020 11:06 pm Those are very interesting. Especially the different shades of greens on Duck Tales (not as NTSC artifacts, but as areas of flat colour).
Reminds me of what we saw as differences between RGB and YUV output from the 6260 on PC Engine over the past several months (as the new palette was developed).

Does the composite adapter use RGB as input or YUV ?
RGB
d909
Posts: 35
Joined: Sat Sep 12, 2020 3:21 am
Has thanked: 1 time
Been thanked: 10 times

Re: Composite Adapter for MiSTer

Unread post by d909 »

dshadoff wrote: Fri Oct 09, 2020 11:06 pm Those are very interesting. Especially the different shades of greens on Duck Tales (not as NTSC artifacts, but as areas of flat colour).
Reminds me of what we saw as differences between RGB and YUV output from the 6260 on PC Engine over the past several months (as the new palette was developed).

Does the composite adapter use RGB as input or YUV ?
Don't read too much in to the different color shades. NTSC signal colors are not as rigidly defined as RGB. There are hue and chroma knobs on many displays that only work with such signals, and not RGB. The difference you're seeing could simply be due to adjustments on my monitor.
dshadoff
Core Developer
Posts: 534
Joined: Sun May 24, 2020 9:30 pm
Has thanked: 19 times
Been thanked: 141 times

Re: Composite Adapter for MiSTer

Unread post by dshadoff »

If I hadn't been staring at these specifically over the past 5 months working with others to correct PC Engine output, I'd probably agree.

My point being, however, that you might want to play PC Engine games using the "RGB" colours instead of the "original" colors - if the adapter does a RGB-YUV transformation. We encoded a RGB-YUV-RGB transformation into the core for "original".

I just purchased one of these because you have sufficiently piqued my curiosity.
(this is in addition to another different composite adaptor I also have on order...)
d909
Posts: 35
Joined: Sat Sep 12, 2020 3:21 am
Has thanked: 1 time
Been thanked: 10 times

Re: Composite Adapter for MiSTer

Unread post by d909 »

dshadoff wrote: Fri Oct 09, 2020 11:15 pm My point being, however, that you might want to play PC Engine games using the "RGB" colours instead of the "original" colors - if the adapter does a RGB-YUV transformation. We encoded a RGB-YUV-RGB transformation into the core for "original".
Oh, interesting. I see.
User avatar
aberu
Core Developer
Posts: 1144
Joined: Tue Jun 09, 2020 8:34 pm
Location: Longmont, CO
Has thanked: 244 times
Been thanked: 388 times
Contact:

Re: Composite Adapter for MiSTer

Unread post by aberu »

Newsdee wrote: Tue Sep 22, 2020 9:23 am As far as I know there is no adapter that will produce perfect composite with MiSTer.

You can't fault antonio for trying to support multiple FPGA boards, propose a solution to MiSTer that works, and getting turned down. Different people will have different preferences. Maybe once more adapters get made there will be an accepted common solution in the framework.
I don't think anyone is faulting antonio for trying stuff out and working hard to come up with some adapters and solutions.

Instead, I think people, like me and others, who bought his product are upset for not being informed up front that there was a significantly jarring problem with color artifacts (especially on brighter colors like white) that is inherent to any MiSTer composite adapter... something antonio KNEW about since October 11, 2019 (https://www.atari-forum.com/viewtopic.php?t=37611), and he has not informed his customers beforehand with any sort of warning. Additionally, when I asked him about this issue I was seeing in hopes of fixing it, he emailed me that same link above, he threw Sorgelig under the bus, and told me I would "unfortunately" have to buy his new adapter, which is very unacceptable behavior in my opinion. Needless to say after feeling like I got baited and switched when he's known for a long time this is a problem and he has a solution in the works... I will not be buying another one.

For reference I have tested this on 3 different CRTs. People can post all these PVM screenshots on other $2000 20 inch CRTs, or $500 12 inch CRT's all they want, it is more apparent in motion with your own eyes. I compared the Genesis core to a JVC X'eye I have on all 3 CRT's because I wanted to rule out problems between models. I have a JVC, Mitsubishi, and a more modern Sony Trinitron Flat CRT (which has similar hardware to a PVM). All 3 produced the same exact result.

It would be nice to see antonio go "oh there is a problem" once he was informed of it and just put a red disclaimer about the composite generating *moving* color artifacts on the screen, way back in October 11, 2019, and keep it there. But instead he silently knows about it, sells lots of these adapters, comes here and denies there is a problem (because I guess semantically he's correct as there isn't a problem *with his adapters* per se if we assume he's correct that it's entirely on FPGA dev to fix the problem, which I'm inclined not to believe), and then tells people that email him for support to "buy another one" whenever that comes out. It's just bad customer service.
antoniovillena wrote: Fri Sep 04, 2020 8:16 am
retrorepair wrote: Wed Sep 02, 2020 3:25 pm I would like to see an example of this if someone can show me?

I have a feeling the issue has been somewhat over-hyped, I have not noticed the problem on my own adapter, though admittedly it's using a different encoder IC.
Yes. This guy is seller of MiSTer stuff and it seems he does to attack competitors. As I explain on that email, the problem is due to the core and it's present on all composite adapters on the market not only my one. I won't say me adapter it's the best, but I just receive this email from a client.
Excuse me. That is my twitter post. I am not a seller of MiSTer stuff and never have been, and I'm not attacking competitors. Can you stop lying? Thanks.
birdybro~
retrorepair
Posts: 257
Joined: Sun May 24, 2020 9:06 pm
Has thanked: 64 times
Been thanked: 13 times

Re: Composite Adapter for MiSTer

Unread post by retrorepair »

aberu wrote: Tue Oct 20, 2020 3:38 pm
Newsdee wrote: Tue Sep 22, 2020 9:23 am As far as I know there is no adapter that will produce perfect composite with MiSTer.

You can't fault antonio for trying to support multiple FPGA boards, propose a solution to MiSTer that works, and getting turned down. Different people will have different preferences. Maybe once more adapters get made there will be an accepted common solution in the framework.
I don't think anyone is faulting antonio for trying stuff out and working hard to come up with some adapters and solutions.

Instead, I think people, like me and others, who bought his product are upset for not being informed up front that there was a significantly jarring problem with color artifacts (especially on brighter colors like white) that is inherent to any MiSTer composite adapter... something antonio KNEW about since October 11, 2019 (https://www.atari-forum.com/viewtopic.php?t=37611), and he has not informed his customers beforehand with any sort of warning. Additionally, when I asked him about this issue I was seeing in hopes of fixing it, he emailed me that same link above, he threw Sorgelig under the bus, and told me I would "unfortunately" have to buy his new adapter, which is very unacceptable behavior in my opinion. Needless to say after feeling like I got baited and switched when he's known for a long time this is a problem and he has a solution in the works... I will not be buying another one.

For reference I have tested this on 3 different CRTs. People can post all these PVM screenshots on other $2000 20 inch CRTs, or $500 12 inch CRT's all they want, it is more apparent in motion with your own eyes. I compared the Genesis core to a JVC X'eye I have on all 3 CRT's because I wanted to rule out problems between models. I have a JVC, Mitsubishi, and a more modern Sony Trinitron Flat CRT (which has similar hardware to a PVM). All 3 produced the same exact result.

It would be nice to see antonio go "oh there is a problem" once he was informed of it and just put a red disclaimer about the composite generating *moving* color artifacts on the screen, way back in October 11, 2019, and keep it there. But instead he silently knows about it, sells lots of these adapters, comes here and denies there is a problem (because I guess semantically he's correct as there isn't a problem *with his adapters* per se), and then tells people that email him for support to "buy another one" whenever that comes out. It's just bad customer service.
antoniovillena wrote: Fri Sep 04, 2020 8:16 am
retrorepair wrote: Wed Sep 02, 2020 3:25 pm I would like to see an example of this if someone can show me?

I have a feeling the issue has been somewhat over-hyped, I have not noticed the problem on my own adapter, though admittedly it's using a different encoder IC.
Yes. This guy is seller of MiSTer stuff and it seems he does to attack competitors. As I explain on that email, the problem is due to the core and it's present on all composite adapters on the market not only my one. I won't say me adapter it's the best, but I just receive this email from a client.
Excuse me. That is my twitter post. I am not a seller of MiSTer stuff and never have been, and I'm not attacking competitors. Can you stop lying? Thanks.
The trouble is, it's not an issue that's really anything to do with MiSTer or Antonio's PCBs. It's an inherent problem with composite video generation AND how it's decoded on the display your using.

You may find on a TV from the mid 80s, it's hardly noticeable at all because the TV expects the signal to do this and compensates for it but a more modern set will likely show it up more as it tries to sharpen composite video (see "trying to shine a turd").

Some hardware manufacturers found a way round it by using a clock they already had on the system to generate a subcarrier close enough to the PAL or NTSC standards to work. A hell of a lot didn't though.

It's a known composite video trait so I think attacking him is ridiculous.
User avatar
aberu
Core Developer
Posts: 1144
Joined: Tue Jun 09, 2020 8:34 pm
Location: Longmont, CO
Has thanked: 244 times
Been thanked: 388 times
Contact:

Re: Composite Adapter for MiSTer

Unread post by aberu »

retrorepair wrote: Tue Oct 20, 2020 3:54 pm
The trouble is, it's not an issue that's really anything to do with MiSTer or Antonio's PCBs. It's an inherent problem with composite video generation AND how it's decoded on the display your using.

You may find on a TV from the mid 80s, it's hardly noticeable at all because the TV expects the signal to do this and compensates for it but a more modern set will likely show it up more as it tries to sharpen composite video (see "trying to shine a turd").

Some hardware manufacturers found a way round it by using a clock they already had on the system to generate a subcarrier close enough to the PAL or NTSC standards to work. A hell of a lot didn't though.

It's a known composite video trait so I think attacking him is ridiculous.
If it's a known problem that can't be fixed, then:

1. Why does he say it will be fixed in the next version and tell me to buy that one next when it comes out?

2. Why does he lie about me being a competitor and attacking him for that motivation (implying I'm lying about the literal email he sent me in the twitter post)?

3. Why does the problem not occur with my JVC X'eye on 3 different CRT's that were made in the early 90's, late 90's and early 2000's from three different brands? Why did I never see this across my entire childhood across multiple consoles?

I am not asserting that he ought to fix this now or refund anyone or anything like that. I'm saying he knows this is a problem, plenty of people have emailed him about it, and he practically has a canned response for it now, and he is still selling composite adapters knowing people will continue to be dissatisfied.
birdybro~
retrorepair
Posts: 257
Joined: Sun May 24, 2020 9:06 pm
Has thanked: 64 times
Been thanked: 13 times

Re: Composite Adapter for MiSTer

Unread post by retrorepair »

aberu wrote: Tue Oct 20, 2020 3:59 pm
If it's a known problem that can't be fixed, then:

1. Why does he say it will be fixed in the next version and tell me to buy that one next when it comes out?
As I said above, it can be fixed, but requires a special build of MiSTer as Sorg refuses to allow it to the main build. Again, it's not a fault of the design!
aberu wrote: Tue Oct 20, 2020 3:59 pm
3. Why does the problem not occur with my JVC X'eye on 3 different CRT's that were made in the early 90's, late 90's and early 2000's from three different brands? Why did I never see this across my entire childhood across multiple consoles?
Because your console and a lot of the consoles you have used have a built in work around to fix the issue, much the same as how Antonio now has.

I'll bet you all the tea in China though, you HAVE seen it before but never noticed it till now.

Hook up a DVD player, laserdisc player or VCR, high chance you'll see it. I'd bet Atari 2600 has it too, though I've not one here to test.
retrorepair
Posts: 257
Joined: Sun May 24, 2020 9:06 pm
Has thanked: 64 times
Been thanked: 13 times

Re: Composite Adapter for MiSTer

Unread post by retrorepair »

User avatar
aberu
Core Developer
Posts: 1144
Joined: Tue Jun 09, 2020 8:34 pm
Location: Longmont, CO
Has thanked: 244 times
Been thanked: 388 times
Contact:

Re: Composite Adapter for MiSTer

Unread post by aberu »

retrorepair wrote: Tue Oct 20, 2020 4:07 pm
As I said above, it can be fixed, but requires a special build of MiSTer as Sorg refuses to allow it to the main build. Again, it's not a fault of the design!
This doesn't address the fact that antonio knows that this happens, knows that people don't expect it to happen, knows it looks different than original hardware, advertises that his adapter will output just like original hardware on his page, and then proceeds to try and sell us another one after we ask him how to fix the problem. My main point is still... why doesn't he inform people of this up front since he's known about the problem for literally over a year now? Why does he still say his adapter outputs just like original hardware when it actually doesn't?
retrorepair wrote: Tue Oct 20, 2020 4:07 pm
Because your console and a lot of the consoles you have used have a built in work around to fix the issue, much the same as how Antonio now has.

I'll bet you all the tea in China though, you HAVE seen it before but never noticed it till now.

Hook up a DVD player, laserdisc player or VCR, high chance you'll see it. I'd bet Atari 2600 has it too, though I've not one here to test.
I don't think you know what I mean. It's like you load up a still image, like a menu in an RPG. It has white borders. Those white borders with his adapter basically have a rainbow moving over them. That rainbow moves faster or slower if you move the VGA jumper btw. My first thought was misconfiguration. I changed the jumper for the VGA on the analog i/o board, that just changed the speed of the rainbow somewhat. I tried different CRTs, luggin them around. I tried different composite cables just in case. I tried to isolate for EMI reasons. I even pulled out a power conditioner I have and used that to rule out the power being too dirty to my CRTs. Same thing. I then pulled out my JVC X'Eye and compared same games (Genesis Shining Force 2 and Phantasy Star IV for the Genesis core from my collection, and Shining Force CD on Sega CD). Everything was as expected, the color artifacts don't move from what I can tell at all.

On the MiSTer cores the same color artifacts happened in the SNES core, the TGFX16 core, the Genesis core, Sega CD core. Do you have other ones you want me to test tonight? Any advice on how I can take a video of the issue to reliably convey what it is? I've tried to but it always comes up looking bad since recording CRTs is tough. Thanks.

As for question 2, again, why do you think he lied about me and said I'm a competitor? Just a thought.

EDIT: Whatever, last thing I'm saying here:

I am not attacking antoniovillena's work as a whole. Antonio certainly makes some great products. This is about the composite adapter (which is a fine s-video adapter btw) debacle. I'm frustrated by his repeated response to this for quite some time being wholly inadequate. My advice to antonio is; Don't sell things to people with false pretenses, don't lie about anyone who criticizes your work, and don't throw Sorgelig under the bus unnecessarily as a shield for criticism, when Sorgelig was a huge part of creating the opportunity for you to make money off of the MiSTer project in the first place.

EDIT2: "Sorg refuses to allow it to the main build" <--- Please present evidence of the pull requests that were denied when you claim things like this.
birdybro~
gordonfish
Posts: 9
Joined: Thu Nov 19, 2020 1:04 am
Has thanked: 21 times
Been thanked: 2 times

Re: Composite Adapter for MiSTer

Unread post by gordonfish »

Been reading through this thread, and while I understand the desire to have absolutely perfect-like-original output, it really feels to me like people really forget how inconsistent composite (as well as S-video and esepcially RF) could be.

I remember back in the 90s of doing an experiment with a couple of friends, who each brought their NES (USA front loaders, all the same model) that were each connected via composite one at a time, by quickly swapping the same RCA video cable between them. The point of the experiment was to see if they all ran at the same speed, to see if they all looked the same, and what, if any, differences there were.

What I recall is each NES's composite output did vary slightly in color, and one of the NES units was visibly darker (not by a whole lot) than the other two. Artifacting also varied a from one NES to the others. The text in the SMB1 title screen I recall looked a little different between the systems, on at least one of the units looking slightly smoother than a third one.

The real take away here is that it was never the exact same image. I believe that people today have gotten so used to having a consistent video picture on HD and 4K televisions, that many of us forget how analog output generation, as well as CRT displays, could vary quite a lot.

So I humbly submit that if the output can vary from one genuine NES to another, than seeing slight differences between a real NES and the composite adapter feels more inline with the reality of analog output generation. I agree there could always be room for improvement (like what was said earlier in this tread about better syncing once cores get updated with that ability), though currently it looks like we can get really close with some of the adapters posted here and elsewhere.

The quest for 100% consistency where it didn't exist between real consoles, isn't a realistic goal, imho, especially when you factor in 1) all the different composite generation hardware used in different game and computer consoles that used it, as well as 2) runs and revisions of the same consoles could have used varying components for analog video generation, giving way to even more potential inconsistency between units of the same console.
dshadoff
Core Developer
Posts: 534
Joined: Sun May 24, 2020 9:30 pm
Has thanked: 19 times
Been thanked: 141 times

Re: Composite Adapter for MiSTer

Unread post by dshadoff »

This was basically the point that I started with as well, being a person who experienced composite for over half of my life.
So I purchased one of Antonio's adapters to see for myself.

Unfortunately, I haven't got it to work reliably yet. I get a picture for about a second, then nothing (and repeat).
LeftEmpty
Posts: 141
Joined: Sun May 24, 2020 6:47 pm
Has thanked: 2 times
Been thanked: 4 times

Re: Composite Adapter for MiSTer

Unread post by LeftEmpty »

I bought an RGB to composite adapter from Axunworks and it has been interesting, but I think I should test it further.

The adapter worked only through VGA to VGA connection through the I/O board (using a VGA to SCART adapter didn't work at all). Only a few cores actually worked: the NES, SNES and PCE ones mainly. The Megadrive one didn't at all.
The artefacting was really intense, on par with the only real composite hardware I have left, which is for the PC Engine. It seemed to be much stronger than what I remembered from the NES and SNES composite.
As expected, this required quite a lot of tunage versus standard RGB, which means it's really more of a curiosity I'd use on occasion than something I'd do regularly, because of all the hassle of having to fiddle with cables first, and then tune the CRT, and back again for other RGB consoles.
User avatar
aberu
Core Developer
Posts: 1144
Joined: Tue Jun 09, 2020 8:34 pm
Location: Longmont, CO
Has thanked: 244 times
Been thanked: 388 times
Contact:

Re: Composite Adapter for MiSTer

Unread post by aberu »

gordonfish wrote: Fri Nov 27, 2020 10:09 pm Been reading through this thread, and while I understand the desire to have absolutely perfect-like-original output, it really feels to me like people really forget how inconsistent composite (as well as S-video and esepcially RF) could be.

I remember back in the 90s of doing an experiment with a couple of friends, who each brought their NES (USA front loaders, all the same model) that were each connected via composite one at a time, by quickly swapping the same RCA video cable between them. The point of the experiment was to see if they all ran at the same speed, to see if they all looked the same, and what, if any, differences there were.

What I recall is each NES's composite output did vary slightly in color, and one of the NES units was visibly darker (not by a whole lot) than the other two. Artifacting also varied a from one NES to the others. The text in the SMB1 title screen I recall looked a little different between the systems, on at least one of the units looking slightly smoother than a third one.

The real take away here is that it was never the exact same image. I believe that people today have gotten so used to having a consistent video picture on HD and 4K televisions, that many of us forget how analog output generation, as well as CRT displays, could vary quite a lot.

So I humbly submit that if the output can vary from one genuine NES to another, than seeing slight differences between a real NES and the composite adapter feels more inline with the reality of analog output generation. I agree there could always be room for improvement (like what was said earlier in this tread about better syncing once cores get updated with that ability), though currently it looks like we can get really close with some of the adapters posted here and elsewhere.

The quest for 100% consistency where it didn't exist between real consoles, isn't a realistic goal, imho, especially when you factor in 1) all the different composite generation hardware used in different game and computer consoles that used it, as well as 2) runs and revisions of the same consoles could have used varying components for analog video generation, giving way to even more potential inconsistency between units of the same console.
I gotta come back for this since I feel like I'm being literally gaslighted on this, and people are incorrectly being led to believe I'm delusional or something or just misremembering. This is not a "slight difference", this is a directly demonstrable severe difference. I can't test it anymore since I literally gave antonio's adapter away for free since I didn't feel it was right to sell it for anything since I knew of it's significant flaws. Also, unlike antonio, I actively informed the person I gave it away to for free that it had significant issues. Because that's what people with some shred of *integrity* do.

I have now tested the same games on a Sega Mega Jet, Genesis Model 1 (VA6), and a JVC X'eye, and none of them exhibit the same behavior, across all 3 CRT's that span over a decade of manufacturing and 3 different brands. I have to say that it feels pretty frustrating to be told that I have false memories when I have done direct live comparisons with original hardware in person. Again, let's be clear, this is not dot crawl. This was rainbow luma interference, as described here --> https://www.renesas.com/us/en/document/ ... techniques
colortrap.png
colortrap.png (85.9 KiB) Viewed 7336 times
Colors from the luma signal are bleeding into the composite signal, because either there is now YTRAP or there is an insufficient YTRAP, is the most likely cause, but whatever.

If you are thinking that white borders around characters avatars in games like Shining Force II had a constantly moving wave of rainbows flying over them, I gotta dispute that, since that effect literally didn't happen across 3 different systems now on 3 different CRT's, for the same platform (Sega Genesis) with the original hardware I have on hand. The only time that happened was with antoniovillena's composite adapter. The only time I saw this super jarring effect that generates problems. Maybe you all are playing on PAL CRT's, I'm not.I switched his to NTSC, I tried different jumpers, I tried multiple permutations of settings combinations, I tried removing the jumper (obv didn't work), so many more things as I've already said here.The frequency of the Rainbow chroma effect changed with 3.3v vs 5v on the vga jumper.

Fact is if you say "it generates the same signal as original hardware", you'd better be right, since it's kinda insulting when it obviously does not. I demonstrated it doesn't, dshadoff has also seen significant deviation from original hardware and they are an expert on the PC-Engine video output for the core development.
birdybro~
dshadoff
Core Developer
Posts: 534
Joined: Sun May 24, 2020 9:30 pm
Has thanked: 19 times
Been thanked: 141 times

Re: Composite Adapter for MiSTer

Unread post by dshadoff »

To be clear, I haven't got my AntonioVillena adapter working to the point where I can compare the output, but my intention was to evaluate the output.
User avatar
aberu
Core Developer
Posts: 1144
Joined: Tue Jun 09, 2020 8:34 pm
Location: Longmont, CO
Has thanked: 244 times
Been thanked: 388 times
Contact:

Re: Composite Adapter for MiSTer

Unread post by aberu »

Right. If it were generating the same signals, then it would work on your CRT correct? :P
birdybro~
gordonfish
Posts: 9
Joined: Thu Nov 19, 2020 1:04 am
Has thanked: 21 times
Been thanked: 2 times

Re: Composite Adapter for MiSTer

Unread post by gordonfish »

aberu wrote: Tue Dec 08, 2020 4:22 am I gotta come back for this since I feel like I'm being literally gaslighted on this, and people are incorrectly being led to believe I'm delusional or something or just misremembering.
I never intended to imply or give that idea to anyone, and I apologize if I came across that way, as that's not how I feel.

I don't have the adapter in question so I cannot speak to what it actually produces.

My main point was simply that composite (and RF as well as S-Video, and even to some extent, component) was/is not consistent in the same way HDMI, DVI, and DP are, in that you could take multiple instances of the same system set to the same resolution and showing the same image, connected through an HDMI, DP, etc switcher and observe the same pixel-consistent output for each of them (barring LCD scaling shenanigans), where as the old analog connections, as well as analog TVs/monitors (or ports, as I had a GE 27" tv circa 1990 where one RCA input looked better than the other) could indeed vary quite a bit (producing different artifacts or to different degrees, that sort of thing.)
This is not a "slight difference", this is a directly demonstrable severe difference.
I have now tested the same games on a Sega Mega Jet, Genesis Model 1 (VA6), and a JVC X'eye, and none of them exhibit the same behavior, across all 3 CRT's that span over a decade of manufacturing and 3 different brands.
I don't doubt any of what you're saying, as I don't have this adapter. Though one thing I wonder is if that's enough of a sample pool, if only because I remember seeing a Genesis model 2 in the 1990s that had some strange composite artifacts that didn't occur over RF on the same unit, though I can't be sure it wasn't the TV (I don't believe it was the cable since it was also used with my Genesis model 1, but since an adapter had to be used for the different size DIN on the model 2, that might have been the source of the problem too.)
I have to say that it feels pretty frustrating to be told that I have false memories
Again, I never meant to imply anything like that, and feel that is awful to do to anyone.
when I have done direct live comparisons with original hardware in person.
I too had done tests, with NESes as I described in my post, where there were some very noticeable inconsistencies, though I can't say if they were at the same level as what you saw with the adapter in question. I can only say that there was enough inconsistency to make me not want to rush to conclusions about things like adapters in general. I believe you are in earnest about you saw and I don't doubt your tests, though it's be nice to see a video showing what you've observed, if only to get a better idea of what you're were seeing.
Again, let's be clear, this is not dot crawl. This was rainbow luma interference, as described here --> https://www.renesas.com/us/en/document/ ... techniques
Thanks, I will take a look at this.
Fact is if you say "it generates the same signal as original hardware", you'd better be right, since it's kinda insulting when it obviously does not.
I agree completely. It should at least be very close, within the same margin of error that can be demonstrated on an array of real hardware. We just can't and should not expect 100% consistency with one or two particular units of an original console, but need to be aware of how various units of one console could actually vary (like in that NES test I described in my post.)
User avatar
aberu
Core Developer
Posts: 1144
Joined: Tue Jun 09, 2020 8:34 pm
Location: Longmont, CO
Has thanked: 244 times
Been thanked: 388 times
Contact:

Re: Composite Adapter for MiSTer

Unread post by aberu »

gordonfish wrote: Tue Dec 08, 2020 4:18 pm I never intended to imply or give that idea to anyone, and I apologize if I came across that way, as that's not how I feel.
Sorry for kinda snapping on you... it's been a frustrating subject precisely because antonio has deliberately lied about me and there have been others who are saying I just don't remember what composite was like, which is false, and irrelevant since I directly compared now with multiple pieces of hardware spanning over a decade since being manufactured. You just kinda inadvertently piled on and got caught in the crossfire, I apologize.
gordonfish wrote: Tue Dec 08, 2020 4:18 pmI don't doubt any of what you're saying, as I don't have this adapter. Though one thing I wonder is if that's enough of a sample pool, if only because I remember seeing a Genesis model 2 in the 1990s that had some strange composite artifacts that didn't occur over RF on the same unit, though I can't be sure it wasn't the TV (I don't believe it was the cable since it was also used with my Genesis model 1, but since an adapter had to be used for the different size DIN on the model 2, that might have been the source of the problem too.)
If 3 CRTs from 3 brands across 12+ years of MFG, with 3 different models of essentially the same system attached to each (9 permutations) isn't enough samples to confirm how original hardware works on CRTs in general, then I think your bar for what is a good sample size might be a bit unreasonable personally, for the average consumer. That's the reference sample size of 3x3, the comparison was one MiSTer on 3 different CRT's all showing the same problem when compared to the reference...
gordonfish wrote: Tue Dec 08, 2020 4:18 pmI agree completely. It should at least be very close, within the same margin of error that can be demonstrated on an array of real hardware. We just can't and should not expect 100% consistency with one or two particular units of an original console, but need to be aware of how various units of one console could actually vary (like in that NES test I described in my post.)
I never expected 100%, but the color rainbow spamming across all bright colors that was more pronounced the brighter the color, was incredibly jarring and distracting and made it basically unusable. I'm not *that* picky :P
birdybro~
HappehLemons
Posts: 28
Joined: Wed Oct 13, 2021 3:17 am
Has thanked: 3 times
Been thanked: 11 times

Re: Composite Adapter for MiSTer

Unread post by HappehLemons »

Bump of the dead here, but I am having similar issues and wondering if anyone has found a solution.

It's hard to figure out if this is due to Antonio's adapter though, because while I do notice this on NES & SNES, it's something I have to look for to see.

However, the Genesis core looks truly awful. Intense rainbow banding, dot crawl and colors that look way off. Even my wife who see me play retro games all the time but doesn't *look* for these issues asked me "What's wrong with the video on this?".

Compared to my original Genesis model 2, I get no rainbow banding, dot crawling and color that look correct. Castlevaina is a prime example, although I didn't take this picture myself, it looks very close to the Analogue DAC genesis video bug pictured below.


My next CRT may not have component inputs, so I was looking for a composite solution (Also I prefer it for Genesis due to the dithering) .
Attachments
Capture.JPG
Capture.JPG (20.13 KiB) Viewed 4936 times
Post Reply