• Hey everyone, staff have documented a list of banned content and subject matter that we feel are not consistent with site values, and don't make sense to host discussion of on Famiboards. This list (and the relevant reasoning per item) is viewable here.
  • Do you have audio editing experience and want to help out with the Famiboards Discussion Club Podcast? If so, we're looking for help and would love to have you on the team! Just let us know in the Podcast Thread if you are interested!

StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)

Reader mode
Erimgard said:
The specific criticism is about claiming the Pro is coming in 2019, 2020, and 2021, and each time claiming that he got nothing wrong, Nintendo just delayed it again. No one is insulting him for being good at his job. It's fine to be skeptical of his claims.
Click to expand...
He never claimed it was coming in 2020. And the device he claimed was coming in 2019 did come, it just didn't have "enhanced features for enthusiasts" unless you're enthusiastic about battery life.
CM132 said:
This is official confirmation that Pro/2/Wii U 2 (lol) exists.
Click to expand...
No it's not. This method was used specifically for upscaling Super Mario Sunshine in SM3DAS. They may use it for other stuff but there is certainly no indication that they definitely will.
 
CM132 said:
I thought that this is something like Nintendo DLSS.
Click to expand...
It is more or less, but NERD explicitly confirmed that they did use machine learning to upscale videos from Sunshine for the Switch version. NERD is the group who filed this patent, so it's extremely likely that this patent is for the machine learning solution they used for Sunshine.

They could be using it for other games too (IMO it'll probably be used for retro games/NSO games) but so far we have near-confirmation on what it was used for so far, and that had nothing to do with a new Switch.
 
Remember the rule: If you find out about something from a Nintendo patent instead of an announcement, it's because they're not actually using it in the thing they're going to announce.

nintendo-patent-portable1.900x.jpg

CM132 said:
I thought that this is something like Nintendo DLSS.
Click to expand...
The new model doesn't need to use Nintendo DLSS though, it has actual DLSS.
 
Skittzo said:
No it's not. This method was used specifically for upscaling Super Mario Sunshine in SM3DAS. They may use it for other stuff but there is certainly no indication that they definitely will.
Click to expand...
Is it? Based on the abstract, and the information in it, SM3DAS does not really fit the description.

”A computer system is provided for converting images through use of a trained neural network. A source image is divided into blocks and context data is added to each pixel block. The context blocks are split into channels and each channel from the same context block is added to the same activation matrix. The action matrix is then executed against a trained neural network to produce a changed activation matrix. The changed activation matrix is then used to generate a converted image.”



What was done in 3DAS was about Upscaling the rendered cutscenes to a higher resolution.


”NERD’s own Nintendo GameCube emulation technology was used in Super Mario 3D All-Stars to bring Super Mario Sunshine to Nintendo Switch.

One of the biggest challenges was emulating the Nintendo GameCube’s old but powerful MPU (microprocessor) on the Switch’s customized processor: a number of optimization tricks were needed to get the game to run at full speed.

In addition, NERD worked with the Super Mario 3D All-Stars team on several features to give Super Mario Sunshine a modern twist. These include 16:9 HD rendering, updated controller bindings for an optimal Joy-Con experience, and others… The in-game videos were also upgraded to HD using NERD’s own deep learning engine.

Leveraging the similarities between Nintendo GameCube and Wii hardware architectures, NERD also supported Super Mario Galaxy porting efforts by providing graphics and audio emulation technologies.”

This NERD super sampling thing is for real-time, what was done to 3DAS is not in real-time. It was done using a neural network but it isn’t real-time on the switch.


LiC said:
Remember the rule: If you find out about something from a Nintendo patent instead of an announcement, it's because they're not actually using it in the thing they're going to announce.

nintendo-patent-portable1.900x.jpg


The new model doesn't need to use Nintendo DLSS though, it has actual DLSS.
Click to expand...
That assumes Nintendo would even bother ever amounting if they were using it real-time or not.

And how likely do you think that would be?
 
Hey all.

The thread has gotten a little out of control lately. Some of that is fun! The exciting speculation, the rumors, the excellent investigative work! The opening of our community to Chinese fans, and the crew of translators, it's great.

Some of it is less cool. It's gotten heated in here, and not just on the stuff that has historically been the most heated. Several people have publicly opted out, a few have opted out privately.

I'm not the OP or a mod, so I'm just asking as someone who wants this to be as welcoming a place as it can be, and to live up to what the rest of Famiboards has wanted to do from the outset. Can we cut the personal attacks, and snide remarks? There has been a lot of chatter lately that isn't on-topic (which is fun, as long as it doesn't dominate) but is also kinda hostile (which is not!).

Lots of new folks in here as well. Let's not just welcome them, but can the Old Folks set a standard for how we talk? How we engage on these topics?
 
0
ReddDreadtheLead said:
Is it? Based on the abstract, and the information in it, SM3DAS does not really fit the description.

”A computer system is provided for converting images through use of a trained neural network. A source image is divided into blocks and context data is added to each pixel block. The context blocks are split into channels and each channel from the same context block is added to the same activation matrix. The action matrix is then executed against a trained neural network to produce a changed activation matrix. The changed activation matrix is then used to generate a converted image.”



What was done in 3DAS was about Upscaling the rendered cutscenes to a higher resolution.


*”NERD’s own Nintendo GameCube emulation technology was used in Super Mario 3D All-Stars to bring Super Mario Sunshine to Nintendo Switch.

One of the biggest challenges was emulating the Nintendo GameCube’s old but powerful MPU (microprocessor) on the Switch’s customized processor: a number of optimization tricks were needed to get the game to run at full speed.

In addition, NERD worked with the Super Mario 3D All-Stars team on several features to give Super Mario Sunshine a modern twist. These include 16:9 HD rendering, updated controller bindings for an optimal Joy-Con experience, and others… The in-game videos were also upgraded to HD using NERD’s own deep learning engine.

Leveraging the similarities between Nintendo GameCube and Wii hardware architectures, NERD also supported Super Mario Galaxy porting efforts by providing graphics and audio emulation technologies.*”

This NERD super sampling thing is for real-time, what was done to 3DAS is not in real-time. It was done using a neural network but it isn’t real-time on the switch.
Click to expand...
Nothing in claim 1 states that this transform process is happening in real time. The patent specification for sure says it can work in real time, but it can also be done ahead of time like we saw with Sunshine.
 
Erimgard said:
The specific criticism is about claiming the Pro is coming in 2019, 2020, and 2021, and each time claiming that he got nothing wrong, Nintendo just delayed it again. No one is insulting him for being good at his job. It's fine to be skeptical of his claims.
Click to expand...
The only time Mochizuki has claimed a "Pro" was coming is late 2021 before the OLED. Saying he did it in 2020 is simply false, and saying he did it in 2019 is a gross exaggeration. What he actually published in 2019 was a March report about the Lite in WSJ, which also said another model with "advanced features targeted at enthusiast gamers" was coming. That was followed a month later by him co-authoring a report in Bloomberg, which corroborated the Lite, but specifically said the other model was a "modest upgrade" and wasn't more powerful. That all turned out to be quite accurate! And we all know what happened with the OLED.
 
LiC said:
The only time Mochizuki has claimed a "Pro" was coming is late 2021 before the OLED. Saying he did it in 2020 is simply false, and saying he did it in 2019 is a gross exaggeration. What he actually published in 2019 was a March report about the Lite in WSJ, which also said another model with "advanced features targeting at enthusiast gamers" was coming. That was followed a month later by him co-authoring a report in Bloomberg which corroborated the Lite but specifically said the other model was a "modest upgrade" and wasn't more powerful. That all turned out to be quite accurate! And we all know what happened with the OLED.
Click to expand...
Yeah, no one said anything about a new hardware in 2020.
Nintendo said that there was no plan for new hardware for that year.
 
LiC said:
The only time Mochizuki has claimed a "Pro" was coming is late 2021 before the OLED. Saying he did it in 2020 is simply false, and saying he did it in 2019 is a gross exaggeration. What he actually published in 2019 was a March report about the Lite in WSJ, which also said another model with "advanced features targeted at enthusiast gamers" was coming. That was followed a month later by him co-authoring a report in Bloomberg, which corroborated the Lite, but specifically said the other model was a "modest upgrade" and wasn't more powerful. That all turned out to be quite accurate! And we all know what happened with the OLED.
Click to expand...

We seem to agree then. He has information, but mixes those with his own speculation and ends up with moderately (Mariko) to largely inaccurate (OLED) reporting.
He'd be a great insider for famiboards, but we can't hold him in similar standards.
 
Skittzo said:
Nothing in claim 1 states that this transform process is happening in real time. The patent specification for sure says it can work in real time, but it can also be done ahead of time like we saw with Sunshine.
Click to expand...
Perhaps not in claim one, but in the technical overview and summary they use the example of real-time usage for this.

“The technology described herein relates to machine learning and using machine learning to convert one dataset or signal into another dataset or signal. More particularly, the technology described herein relates to applying block transforms to such datasets or signal. Applications of the technology include converting images of one resolution into another (e.g., higher) resolution and may be used in real-time applications from images generated by, for example, a video game engine.”



I’m not sure where you saw ahead of time though, and what happens there wouldn’t be applicable to sunshine because it’s not doing it ahead of time on the system, it’s off the system and it’s done by a different device. It’s packaged into 3DAS as an already upscaled outcome. There is no ahead of time and using AI Upscaling is not a first either.
 
ReddDreadtheLead said:
Perhaps not in claim one, but in the technical overview and summary they use the example of real-time usage for this.

“The technology described herein relates to machine learning and using machine learning to convert one dataset or signal into another dataset or signal. More particularly, the technology described herein relates to applying block transforms to such datasets or signal. Applications of the technology include converting images of one resolution into another (e.g., higher) resolution and may be used in real-time applications from images generated by, for example, a video game engine.”



I’m not sure where you saw ahead of time though, and what happens there wouldn’t be applicable to sunshine because it’s not doing it ahead of time on the system, it’s off the system and it’s done by a different device. It’s packaged into 3DAS as an already upscaled outcome. There is no ahead of time and using AI Upscaling is not a first either.
Click to expand...
That's what I meant by ahead of time, it was done during development of 3DAS to upres those videos, which were then included into the game. It was used as a development tool ahead of time, rather than an in-engine tool in real time.

All I'm saying here is that this patent was made public shortly after NERD explicitly referred to using deep learning to upres those videos. It's extremely unlikely that they didn't use the invention described in this patent for that purpose.
 
Bosintang said:
We seem to agree then. He has information, but mixes those with his own speculation and ends up with moderately (Mariko) to largely inaccurate (OLED) reporting.
He'd be a great insider for famiboards, but we can't hold him in similar standards.
Click to expand...
How many of his articles did you actually read? Honest question.
 
Skittzo said:
That's what I meant by ahead of time, it was done during development of 3DAS to upres those videos, which were then included into the game. It was used as a development tool ahead of time, rather than an in-engine tool in real time.
Click to expand...
But that wouldn’t be what this patent is applicable for. This patent isn’t trying to patent AI Upscaling using a NN that you do on a different system, it’s describing something you do on the same device. Whether in the background or in real-time for Upscaling, the process is meant to be done on the device in question. The Neural Network itself can be updated through the internet to be applied to multiple devices to improve the Upscaling Process.


“….The client computing device includes at least one hardware processor and is configured to execute a neural network based on the received image data to generate a target image. The execution of the neural network on the client device applies separable block transforms to a plurality of activation matrices that each correspond to different blocks of pixel data within the image that is represented by the image data. The target image is generated at the target resolution and output, at the target resolution to the display device for display thereon during gameplay of the first video game.”
 
Erimgard said:
The specific criticism is about claiming the Pro is coming in 2019, 2020, and 2021, and each time claiming that he got nothing wrong, Nintendo just delayed it again. No one is insulting him for being good at his job. It's fine to be skeptical of his claims.
Click to expand...
I understand that, but I also think these forums, social media, and YouTube run with the narratives they create.

He probably misstepped, by throwing dates out, but the fact that this forum topic exists shows that we all still believe his reporting on the 4k Switch Pro/successor, and that dev kits are in developer's hands.
 
0
Skittzo said:
He never claimed it was coming in 2020. And the device he claimed was coming in 2019 did come, it just didn't have "enhanced features for enthusiasts" unless you're enthusiastic about battery life.
Click to expand...
I think it's fair to say that with multiple hardware revisions - slight, moderate, and advanced - all happening within a few years of each other (V2, Lite, OLED, some embodiment of Pro/Super/2), that speculative reporting is gonna be wrong, at times.
 
0
iNess said:
the info I mention above is not speculative, it is evidence, that is if the next revision of Switch goes in as an upgrade from HEG-001. for them to jump from HAC to HEG is a significant leap.
Click to expand...
I think we all know Pro/Super/2 it's coming, in the next two years.
 
0
iNess said:
the info I mention above is not speculative, it is evidence, that is if the next revision of Switch goes in as an upgrade from HEG-001. for them to jump from HAC to HEG is a significant leap. There are reports that Q1(April/May/June) 2022-2023 sales are 30% below projection, and Nintendo told Nikkei that they have supply problem, but at the shareholder meeting they closed with raw material inventories over 66 billion yen as of March 31. If they are not parts for the current Switch, then what are they? Or they no longer want to manufacture so many of the current model, reserving those parts for another model, or those raw materials are actually servers that are not yet in production, there is data that they want to implement cloud computing support. In any case, accounting-wise, everything will be clearer until the end of the fiscal year.
Click to expand...
They can have plenty of one type of parts but lack stock of other parts to produce even more consoles, I’m not saying the increase in raw material is not due to new hardware but that there can be both an increase in raw materials and difficulty to produce consoles at the same time.
 
iNess said:
Tests of a new "Switch Oled" model?

at the FCC:
2016 Swtich V1 reviewed under HAC-001
2019 Switch V2 reviewed under HAC-001
i.e. information was updated with a SoC and Memory upgrade letter therefore CPU Mainboard change.

at NCC (Taiwan):
2021: Switch Oled for development was revised HEG-002.
Why not in the FCC? because in the FCC the user manual is uploaded with the DEVKIT specifications. Therefore so far it is kept secret.

At the FCC:
2021 Switch Oled was reviewed under HEG-001.

Already on January 10, 2022 the internal and external photos were to be published. But if you check they are not enabled, why? maybe they issued a confidentiality letter for 180 days not to be published giving them a modified version of Switch Oled to certify, if it was immediately on July 13 it will be one year since the Switch Oled was reviewed at the FCC so they should announce something before that date, or automatically will be visible the letter of modification of the original model with the details.

Analyzing the HEG codename for the Oled
H=Hybrid
E=EL oled in Japanese
G=??? perhaps a reference to "Gameloop".
Click to expand...
I don’t think these HAC or HEG have a significant meaning like that, though I could be wrong here
 
0
Blue Monty said:
They can have plenty of one type of parts but lack stock of other parts to produce even more consoles, I’m not saying the increase in raw material is not due to new hardware but that there can be both an increase in raw materials and difficulty to produce consoles at the same time.
Click to expand...
I'd imagine they would try to buy in bulk, to help offset any future part shortages, as well. Lessons have likely been learned.
 
0
ReddDreadtheLead said:
But that wouldn’t be what this patent is applicable for. This patent isn’t trying to patent AI Upscaling using a NN that you do on a different system, it’s describing something you do on the same device. Whether in the background or in real-time for Upscaling, the process is meant to be done on the device in question. The Neural Network itself can be updated through the internet to be applied to multiple devices to improve the Upscaling Process.


“….The client computing device includes at least one hardware processor and is configured to execute a neural network based on the received image data to generate a target image. The execution of the neural network on the client device applies separable block transforms to a plurality of activation matrices that each correspond to different blocks of pixel data within the image that is represented by the image data. The target image is generated at the target resolution and output, at the target resolution to the display device for display thereon during gameplay of the first video game.”
Click to expand...
So the way patents work, is that the specification describes a broad generalized application of the invention in question, with all sorts of potential uses and changes and differences and embodiments describing many different ways this core idea can be used.

Then the claims describe this core idea, which is the now legally protected invention. The claims do not require it to be real-time, nor do they require it to be on the same device.
 
Erimgard said:
The specific criticism is about claiming the Pro is coming in 2019, 2020, and 2021, and each time claiming that he got nothing wrong, Nintendo just delayed it again. No one is insulting him for being good at his job. It's fine to be skeptical of his claims.
Click to expand...

This is not a thing that actually happened.

What this is trying to attribute to him, comes from somewhere different. Enthusiast news sites took what he said, and inserted their own speculation and conclusion leaping interpretations to what he actually said, and then those telephone game stories dominoed and snowballed to message boards across the internet.

It's important to try and seperate hard things that are said, from speculation and leaps of logic from secondary sources reporting on it.
 
Skittzo said:
So the way patents work, is that the specification describes a broad generalized application of the invention in question, with all sorts of potential uses and changes and differences and embodiments describing many different ways this core idea can be used.

Then the claims describe this core idea, which is the now legally protected invention. The claims do not require it to be real-time, nor do they require it to be on the same device.
Click to expand...
I’m not sure I follow, even within the claims they describe just the computer device executing the neural etwork for producing the transformed image. That would not be on a separate device:


“17. A computer system comprising: a processing system that includes at least one hardware processor, the processing system configured to: populate an initial activation matrix with a plurality of values that are based on data from a plurality of samples from a source signal; apply, over multiple layers of the neural network, separable block transforms that are based on at least a first learned matrix and a second learned matrix, to an input activation matrix to generate a corresponding output activation matrix, wherein the initial activation matrix for each successive layer is the output activation matrix of a prior layer; and output the output activation matrix of the last layer of the neural network to generate a transformed signal that is based on the output activation matrix of the last layer, wherein the source signal is generated based on execution of a computer application program and the transformed signal is output in connection with execution of the computer application program.

18. The computer system of claim 17, wherein the source signal is a source image and the transformed signal is a transformed image.

19. The computer system of claim 18, wherein the source image is at a first resolution and the transformed image is at a second resolution that is different from the first resolution.

20. The computer system of claim 17, wherein the computer application program is a video game application program and the source signal is generated by a rendering engine of the video game application program.”


Granted this isn’t really different from DLSS, as that can be used for videos and pictures but at the moment it is most famous for its game usage.
 
I wonder if Nintendo have saved up some big 3rd party games to show off against this new model? Like the rumoured Batman collection and Borderlands 3, could these be some of the exclusives that have been rumoured? The studios behind these are likely big enough to have access to dev kits.
 
ReddDreadtheLead said:
I’m not sure I follow, even within the claims they describe just the computer device executing the neural etwork for producing the transformed image. That would not be on a separate device:


“17. A computer system comprising: a processing system that includes at least one hardware processor, the processing system configured to: populate an initial activation matrix with a plurality of values that are based on data from a plurality of samples from a source signal; apply, over multiple layers of the neural network, separable block transforms that are based on at least a first learned matrix and a second learned matrix, to an input activation matrix to generate a corresponding output activation matrix, wherein the initial activation matrix for each successive layer is the output activation matrix of a prior layer; and output the output activation matrix of the last layer of the neural network to generate a transformed signal that is based on the output activation matrix of the last layer, wherein the source signal is generated based on execution of a computer application program and the transformed signal is output in connection with execution of the computer application program.

18. The computer system of claim 17, wherein the source signal is a source image and the transformed signal is a transformed image.

19. The computer system of claim 18, wherein the source image is at a first resolution and the transformed image is at a second resolution that is different from the first resolution.

20. The computer system of claim 17, wherein the computer application program is a video game application program and the source signal is generated by a rendering engine of the video game application program.”


Granted this isn’t really different from DLSS, as that can be used for videos and pictures but at the moment it is most famous for its game usage.
Click to expand...
Sorry I meant not all of the claims require it on the same device. Claim 1 specifically just recites a method of doing this matrix math.
 
Crusters said:
This is not a thing that actually happened.

What this is trying to attribute to him, comes from somewhere different. Enthusiast news sites took what he said, and inserted their own speculation and conclusion leaping interpretations to what he actually said, and then those telephone game stories dominoed and snowballed to message boards across the internet.

It's important to try and seperate hard things that are said, from speculation and leaps of logic from secondary sources reporting on it.
Click to expand...
I did some double-checking and a few things were off, but the timeline goes like this:

-He claimed there would be a Switch with enhanced features targeted at avid gamers releasing in 2019
-He retweeted a Digitimes report about a Switch with a stronger CPU being set for 2020 and included a reminder about his 2019 report
-He claimed in August of 2020 that a 4K Switch was coming in Early 2021
-He claimed in Spring of 2021 that a 4K Switch was going to be announced by E3 to launch later that year
 
ITSMILNER said:
I wonder if Nintendo have saved up some big 3rd party games to show off against this new model? Like the rumoured Batman collection and Borderlands 3, could these be some of the exclusives that have been rumoured? The studios behind these are likely big enough to have access to dev kits.
Click to expand...
Skyrim Anniversary edition was the next gen version on all other platforms, and that's leaked.
 
ITSMILNER said:
I wonder if Nintendo have saved up some big 3rd party games to show off against this new model? Like the rumoured Batman collection and Borderlands 3, could these be some of the exclusives that have been rumoured? The studios behind these are likely big enough to have access to dev kits.
Click to expand...
There's also the next-gen version of RDR2 that's still need to be announced for XBX/PS5.
I wonder if they had a deal with Nintendo to only announce that with their future hardware.
 
Erimgard said:
I did some double-checking and a few things were off, but the timeline goes like this:

-He claimed there would be a Switch with enhanced features targeted at avid gamers releasing in 2019
-He retweeted a Digitimes report about a Switch with a stronger CPU being set for 2020 and included a reminder about his 2019 report
-He claimed in August of 2020 that a 4K Switch was coming in Early 2021
-He claimed in Spring of 2021 that a 4K Switch was going to be announced by E3 to launch later that year
Click to expand...

No, he did not, and the record on this has been set straight by multiple users:

LiC said:
The only time Mochizuki has claimed a "Pro" was coming is late 2021 before the OLED. Saying he did it in 2020 is simply false, and saying he did it in 2019 is a gross exaggeration. What he actually published in 2019 was a March report about the Lite in WSJ, which also said another model with "advanced features targeted at enthusiast gamers" was coming. That was followed a month later by him co-authoring a report in Bloomberg, which corroborated the Lite, but specifically said the other model was a "modest upgrade" and wasn't more powerful. That all turned out to be quite accurate! And we all know what happened with the OLED.
Click to expand...

If you want to continue this, you need to provide links for the burden of proof instead of unsubstantiated claims you expect us to take at face value, so everyone can see where your claims and interpretations are actually coming from and what was actually said.
 
Ultimately Mochizuki was always correct when hardware was coming, just not always correct as to what it was or what it all added up to. In cases like the Lite he was spot on, whereas obviously the OLED and the timing of that threw him and others for a loop. I mean Mariko itself likely caused confusion, because it absolutely could have been used for a modest hardware upgrade. In the end it was used for battery life instead.

I think between the Mariko, sourcing of the OLED screen, and the Drake being in development, all sort of swirling around the same time periods, it's understandable how everyone ended up getting caught out in the lead up to the OLED announcement last year. Especially with dev kits being out there for Drake back as early as 2020.
 
Reader mode
You must log in or register to reply here.
Please read this staff post before posting.

Furthermore, according to this follow-up post, all off-topic chat will be moderated.
Last edited:


Back
Top Bottom