PixelFX Morph

[orange808]

Can we finally get tate? Doesn't sound like Mike is going to add it. That's okay, but it sure would be nice to have

I'd prefer to use one upscaler and be done. Video machines have been fun, but I'll have to downsize my life at some point. .

Frame doubling would also be nice. I don't want to pay the full 16 2/3ms penalty for a full frame buffer. My display can recover some of that time.

[Josh128]

+1 for Tate and frame doubling. If we go down the frame doubling rabbit hole, lets include some kind of optional motion interpolation or BFI to counter the double-strobe illusion.


In regards to Tate, simply rotating the image is not enough, for the true Tate arcade experience you will need a mask effect that mimics a rotated CRT, ie vertical scanlines / grille / mask. RT5X is already capable of this and it looks quite 31KHz authentic on a 1080p screen with a 480p source, so pulling it off with a 15KHz game shouldnt be a problem with a 4K scaler.

https://i.imgur.com/FTgGaff.jpg

[orange808]

In regards to Tate, simply rotating the image is not enough, for the true Tate arcade experience you will need a mask effect that mimics a rotated CRT, ie vertical scanlines / grille / mask. RT5X is already capable of this and it looks quite 31KHz authentic on a 1080p screen with a 480p source, so pulling it off with a 15KHz game shouldnt be a problem with a 4K scaler.

https://i.imgur.com/FTgGaff.jpg

Yes, MiSTer already has rotated effects.

They look ridiculous on a big screen, so I really don't care. More options are always nice, but I get the feeling there are a lot of young people using these effects that didn't spend much time with real CRTs. Furthermore, no manufacturer would have released a theoretical huge tube television to the mass market with masks/grilles that were obvious. It looks fake.

[Josh128]

Im not sure what you are looking at, but if they look ridiculous, its because they dont look authentic. True 240p Tate arcade games like Galaga , Pac Man, and Donkey Kong had obvious vertical scanlines, even on the 19" or whatever sized CRT they were played on. The size of the screen doesnt matter at all. The whole point of these boxes is about displaying games the way they looked when we played them, so I have no idea why anyone would oppose that. These boxes + a 4K set should easily be able to recreate that, while Ive seen nothing on MisTer that even comes close.

[orange808]

Im not sure what you are looking at, but if they look ridiculous, its because they dont look authentic. True 240p Tate arcade games like Galaga , Pac Man, and Donkey Kong had obvious vertical scanlines, even on the 19" or whatever sized CRT they were played on. The size of the screen doesnt matter at all. The whole point of these boxes is about displaying games the way they looked when we played them, so I have no idea why anyone would oppose that.

Yes. Scanlines were always there.

However, the appartus mounted in front of the glass (between the gun and glass) wasn't obvious unless you absolutely buried your head in the cab. Even if they get it right, it will still look silly on a big screen. Would be fun on a 32" or 27" screen, though.

I'm not counting on "easily" reproducing anything. Phosphor glow is still unsolved. Adaptive scanlines were a step in the right direction, but it's still not right. It all looks flat and fake. It's better than the emulator look, but I never think it looks like my CRTs.

[Josh128]

They were on most of the arcades and TVs I played on, and even if they werent to you, they were still there and played a part in the image your brain perceived. Regardless, are you arguing that mask effects are pointless? If thats the case, whats the point of the 4K machines? You could just stick to the 5X and be totally fine.

Mike already has misconvergence nailed down. This looks nice.

https://twitter.com/5Xpixels/status/1687626052573466624

[bobrocks95]

I'd rather compare logic blocks, I/O pins, and other details instead of price. Like you said Guspaz, Efinix is pricing aggressively to gain market share.

Here you go.

We're using the Efinix Trion T120 which is $35. It has 112K logic elements, 320 multipliers, 5.4 Mbit block ram, and a 1GHz DDR2 hard memory controller.

The Intel 10CL120 is ~$200. With 119K logic elements, 288 multipliers, 3.8 Mbit block ram, not sure if it has a DDR2 controller.

The Pixel FX hdmi kits use FPGAs from Intel (PS1D HW1), Xilinx (N64D HW1), and Efinix (DCDigital) with the same FX-Framework FPGA code auto-built for each chip. We get the best performance out of Efinix Trion FPGAs hands down, and their price is great so it was an easy decision to switch entirely to Trion FPGAs.

In the end I don't think it's the FPGA that matters, more what you do with it. The Trion T120 can crunch 4K video no problem with plenty of spare room for future updates.

Thanks Woozle, both look like powerful chips with plenty of headroom. Voultar is already on Twitter emphasizing that the Tink 4K's FPGA costs more but implying it is significantly more powerful. From everything I've read it seems like both scalers will be largely comparable with probably a couple odd specific strengths each. I think both will sell well on their own merits, if everyone can cool it with the insults and assumptions until at least both products are available...

CEC protocol

I believe it is not really an important feature for us retro gamers. a complication which is not really necessary. I mean we got a remote and we will have like very few stuff to control anyway.
However, it is nice that people are requesting such stuff that could potentially drive products to be better.

If I'm understanding things from the PixelFX Discord correctly, CEC is used for things like, with a GEM HDMI mod in your system, reporting to the Morph what game you're currently playing and automatically changing video settings based on that. Which imo would be a huge benefit in automating your setup and spending way less time fiddling with settings and way more time playing actual games (which I'm guilty of doing for sure). If I can customize a profile for every PS1 game I own, turn it on, and everything on the Morph sets itself up automatically, that would be incredible.

[spmbx]

The thing with the "crt-effects" is that they look great in stills on twitter but as soon as i configure them at home and things start actually moving things start falling apart. I'm hoping the BFI and hdr options will make things a lot better with actual movement.

[TooBeaucoup]

The thing with the "crt-effects" is that they look great in stills on twitter but as soon as i configure them at home and things start actually moving things start falling apart. I'm hoping the BFI and hdr options will make things a lot better with actual movement.

What do you mean by falling apart? What model of TV do you use? Scanline overlays shouldn't exactly affect motion, at least not from what I've experienced on my displays. Brightness with scanline filters has certainly been an issue in the past, but the 5X with it's HDR implementation as well as being able to tweak brightness on the different mask settings has pretty much eliminated those issues for most.

[spmbx]

What i mean is the motion clarity and crt brightness just isnt there. While a stillshot make it look convincing, seeing it in motion just doesn't cut it. That's where i hope the BFI/hdr options will start making a difference.
This is on a standard 4k lcd which does support hdr but is not getting triggered in my current setup

[Josh128]

What i mean is the motion clarity and crt brightness just isnt there. While a stillshot make it look convincing, seeing it in motion just doesn't cut it. That's where i hope the BFI/hdr options will start making a difference.
This is on a standard 4k lcd which does support hdr but is not getting triggered in my current setup

Thats most likely the issue, and I was thinking about that yesterday. Even the ultra high end LCD TVs dont have per-pixel lighting and contrast, so I can definitely understand why the mask effects could be lacking on them. Their only true advantage is brightness. OLED, on the other hand, is an emissive display technology like CRT and plasma, has ~10x+ the pixel response time of LCD, and the tiny black pixels that will make up the mask effect will be truly black-- like they are on a CRT.

How, may I ask, have you seen the RT4K mask effects in motion?

[citrus3000psi]

1.) Mike was BSing about the hardware cost of the 4K and is milking the price (which I dont believe).

or

2.) The Morph is constructed of signficantly cheaper (and less capable) hardware.


If it somehow turns out that the hardware is similarly capable as the 4K, PixelFX still will have to match the 4K on firmware / features, which will not be an easy task, given that Mike has a huge head start in this regard, building off his experience of the already amazing feature set of the RT5X.

or

3) The moprh is constructed with significantly cheaper (and more capable) hardware.


or

4) The moprh is constructed with roughly the same as the tink4k (and more capable) hardware.


Im sure if the tink4k had stronger hardware he would let us know the part number. Our best guess is its the cyclone 10 LP 055 or 080. But we all know what he was able to pull off with the 5x. So who knows what he is able to do.

[spmbx]

What i mean is the motion clarity and crt brightness just isnt there. While a stillshot make it look convincing, seeing it in motion just doesn't cut it. That's where i hope the BFI/hdr options will start making a difference.
This is on a standard 4k lcd which does support hdr but is not getting triggered in my current setup

Thats most likely the issue, and I was thinking about that yesterday. Even the ultra high end LCD TVs dont have per-pixel lighting and contrast, so I can definitely understand why the mask effects could be lacking on them. Their only true advantage is brightness. OLED, on the other hand, is an emissive display technology like CRT and plasma, has ~10x+ the pixel response time of LCD, and the tiny black pixels that will make up the mask effect will be truly black-- like they are on a CRT.

How, may I ask, have you seen the RT4K mask effects in motion?

I don't think i'm quite following you?
4k as in a 4k tv. I've tried mask effects on my pc connected to it via retroarch and mame with shaders and overlays via numerous presets but never got good results really

[Kez]

I don't think i'm quite following you?
4k as in a 4k tv. I've tried mask effects on my pc connected to it via retroarch and mame with shaders and overlays via numerous presets but never got good results really

I think they mean - there are issues inherent LCD technology vs alternatives like OLED which have benefits in terms of contrast and motion clarity and can exhibit a more CRT-like effect.

[bobrocks95]

I don't think the response time for LCD's can really pull off the more advanced aperture grill/slot mask simulations, I've always heard they fall apart as soon as something starts moving.

[Konsolkongen]

Does “WebUI system control” mean that it will have to be controlled through a website or app? That would be a huge bummer for me. I much prefer physical real buttons on a remote.

Nope, we have physical buttons in the front. And can be controlled by remote. Also, can be changed through the WebUI. We also have other ways to control it coming soon

Excellent

[Josh128]

1.) Mike was BSing about the hardware cost of the 4K and is milking the price (which I dont believe).

or

2.) The Morph is constructed of signficantly cheaper (and less capable) hardware.


If it somehow turns out that the hardware is similarly capable as the 4K, PixelFX still will have to match the 4K on firmware / features, which will not be an easy task, given that Mike has a huge head start in this regard, building off his experience of the already amazing feature set of the RT5X.

or

3) The moprh is constructed with significantly cheaper (and more capable) hardware.


or

4) The moprh is constructed with roughly the same as the tink4k (and more capable) hardware.


Im sure if the tink4k had stronger hardware he would let us know the part number. Our best guess is its the cyclone 10 LP 055 or 080. But we all know what he was able to pull off with the 5x. So who knows what he is able to do.

Sounds like you are throwing down the gauntlet, I like it. Are you saying your machine can and will go head to head with Tink 4K feature wise?

[fernan1234]

Our best guess is its the cyclone 10 LP 055 or 080.

Officially denied:



The mystery continues!

[spmbx]

I just hope it'll be healthy fun competition, but maybe that's a bit of a pipedream.
Sometimes it feels like you're in a european soccer stadium seeing these products compete, which is a shame.

[Konsolkongen]

I wonder if any of these scalers could possibly solve the biggest issue I have with many 2D titles on modern platforms. They are often very poorly scaled, not only looking bad, but also introducing noticeable shimmering when scrolling. This doesn't even apply to just re-released old games, some brand new titles have this issue too!

I know the majority of players don't care or notice, but I do unfortunately - and it is a big enough problem that I won't pick up certain games with this issue even though I would love to play them.

Would it be possible to rescale the input image? Possibly by telling the scaler what the original resolution of the should be, and by adding a few horisontal and vertical markers to show the scaler which which pixels were different in size. Ideally just a few would be needed and it could presumably figure out the interval between poorly scaled pixels itself?

I can understand if this is a big ask, but it would be my wettest dream-feature in a new scaler

[768peeistrash]

respectfully- does it matter what fpga either the rt4k or morph are packing?

..as long as end results are excellent that's what matters to people, along w/ feature set, support, and price of course- some being more sensitive to that than others



when are we gonna get some Morph video footage / screenshots? many wanna see it

[Blair]

If you already have an OSSC or RT5X, you can apparently snag one of these in the digital in only version for $375 and get some 4K goodness. This is quite a brilliant move that could steal some potential RT4K customers due to pricing concerns.

I see no specs on this yet though, wonder what its capabilities are?

I'm assuming one downside of using the OSSC as the analog digitizer in the chain would be the loss of automatic optimal timings, right? Phase and sampling rate would still need to be manually dialed in from the ossc with the help of a test pattern...

[SavagePencil]

Frame doubling would also be nice. I don't want to pay the full 16 2/3ms penalty for a full frame buffer. My display can recover some of that time.

I think I understand this, but can you elaborate on frame doubling? Does this mean the scaler outputs at 120Hz, thus it can do more of its processing in between the frames provided at 60Hz?

[SGGG2]

Pretty sure because the display is receiving at 60hz signal it applies processing at that frame time (16ms). A 120hz signal would be half that, resulting in less lag.

[Josh128]

Frame doubling would also be nice. I don't want to pay the full 16 2/3ms penalty for a full frame buffer. My display can recover some of that time.

I think I understand this, but can you elaborate on frame doubling? Does this mean the scaler outputs at 120Hz, thus it can do more of its processing in between the frames provided at 60Hz?

Yes, and it can also allow the machine to do black or low brightness frame insertion, frame interpolation, etc.

[Fudoh]

I wonder if any of these scalers could possibly solve the biggest issue...

this was discussed early on when the idea for an OSSC Pro was introduced. I think the main reason behind it was the NES mini, but I don't see why it couldn't apply to any emulated 2D games.

In theory it's such a simply feature that I would expect it to implemented down the road on all the new processors. I imagine you could load a pixel bit"map" (just black and white) into memory that tells the processor which pixels of the source to use and which ones to discard. This way a database of maps can be build. The remaining pixels would be used as a new source and upscaled accordingly.

My wishlist would expand beyond this to a version where the processor tries to auto-detect which pixels are original and which ones are doubled and can be discarded. A bit like the phase detection for VGA signals on digital displays.

(EDIT: 13,000th posting)

[bobrocks95]

Since Mike seemed insulted by the suggestion he was using a ~$200 10CL120, maybe he threw in the ~$350 10CX220? That would certainly explain the pricing gap better.

[VEGETA]

they will start a war over fpga choices now, especially that the morph till now looks superior to the tink and cheaper. wonder how stuff will be.

anyway, i was reading about morph feature and got my attention stuff like VRR. is it variable refresh rate? how is that implemented here on games which are 60hz originally? like, how the output changes its fps and isn't this gonna adjust how the game works and feels? for what benefit?

plus, the customized crt masks... are they images stored on sd card where users create them based on their liking then gets overlayed on top of the video? like retroarch masks, or just an image containing the black grid of the mask (AG, slot, shadow...etc) with some faint rgb dots or stripes... then gets overlayed on top. I hope some devs inform me on how it is implemented, will help me understand

over the days I changed my mind about the morph, it is now for me the best option of the 3. all of this for 500$ is nice bargain

[Kez]

VRR. is it variable refresh rate? how is that implemented here on games which are 60hz originally? like, how the output changes its fps and isn't this gonna adjust how the game works and feels? for what benefit?

Classic games are not really VRR, but they often have dated/off-spec refresh rates like 59.82hz and 60.03hz, or even some arcade boards with even wilder refresh rates, and computers at like 70hz. So retro devices like the Tink5x and the MiSTer use the VRR flag on compatible displays not for a variable rate but so that the display will accept the non-standard refresh rates with no screen tearing/stuttering.

[bobrocks95]

they will start a war over fpga choices now, especially that the morph till now looks superior to the tink and cheaper. wonder how stuff will be.

anyway, i was reading about morph feature and got my attention stuff like VRR. is it variable refresh rate? how is that implemented here on games which are 60hz originally? like, how the output changes its fps and isn't this gonna adjust how the game works and feels? for what benefit?

plus, the customized crt masks... are they images stored on sd card where users create them based on their liking then gets overlayed on top of the video? like retroarch masks, or just an image containing the black grid of the mask (AG, slot, shadow...etc) with some faint rgb dots or stripes... then gets overlayed on top. I hope some devs inform me on how it is implemented, will help me understand

over the days I changed my mind about the morph, it is now for me the best option of the 3. all of this for 500$ is nice bargain

Features definitely seem comparable to me right now. The automation on the Morph and integration with their upcoming Infinity Switch sounds like it may win me over, but I can't imagine anyone not wanting to keep an eye on both and see how things play out. This is the competition people were dreaming of when the Morph and OSSC Pro were announced around the same time, and now some are oddly dismissive of competition...

For VRR, I'm not entirely sure what this distinction means but I thought I'd share: