SMWE - New Alternative To HD-CVI, HD-TVI And AHD

Hey Daniel, welcome back!

While we digest that SMWE link, I was wondering if you wanted to give an off the cuff opinion on the feasibility of the AHD, CVI, and TVI combo cameras out there, that apparently support all three transmission types with a single chip, which does not have the blessing any of the three owners of the respective technologies, to boot.

It is also as cheap, if not cheaper than the single transmission ones. In the review that IPVM did their quality fell short of the authorized versions. Although, IMHO, it was respectable for a first build. After all look how far AHD, CVI and TVI have come since their first runs.

The question is I guess, how close are these technologies? Is it just a matter of time before the unauthorized can meet the quality of the authorized units? Or is there some fundamental issue that will prevent a single chip soloution?

Techpoint, Dahua and Nextchip will no doubt pursue any legal avenue to control the rogue maker, but do you think that they could really, even in Dahua's case, be able to stop them, considering the prior art out there?

-Thanks.

Daniel, that's perfectly fine for you to start this. As long as you provide details, it is not an advertisement.

You mention:

"we can avoid the use of an ISP in the camera or extend recording time by a factor of more than 2x (for the same bit rate and image quality)."

Questions:

• How does the cost / price of your components compare to HD analog offerings? Does eliminating an ISP in the camera make much of a difference cost wise?
• How do you extend recording time by a factor of 2? How do you know it is not 10% or 300%?
• What other advantages do you claim for SMWE over HD analog?
• In the future, what resolutions can you support beyond 1080p?

Daniel, I have not read the specs yet, but I do have some prelimary market questions (serious ones). With the three types already available to the market, what makes you unique? Is there a serious need to further fracture a smallish market? Will it require its own formatted recording device? Will it (your recording device) be capable of incorporating any other type of camera at all? I have had discussions recently with some larger manuafacturers that centered around the three types of HD available. My biggest question to them is this: How long will the majority of the market continue to ignore HD Analog? I can find plenty of recording devices/software that can accomodate full resolution of analog and/or IP, but no configuration of HD. Clearly they are hoping it will just magically go away, or that IP camera prices will continue to fall until they drive HD Analog off the island. And yes, I will read the specs.

FWIW, this is not analog, it's digital.

Another thing, Daniel is claiming that he can send RAW sensor data from the sensor chip itself without the need (at the sensor) of an image processing chip.

Why Rockoff didn't find Daniel 7 years ago I'm not sure, but SDI with 1000' range would have been a lot more successful.

Thank-you all for the comments so far. Let me try and answer the more generic points first and later I will answer the specific technical points.

Coming from a broadcast/industrial background I hate it when we deviate from standards. Unless it is for a completely closed system the lack of plug and play capability is a headache for manufacturers and customers alike. Think of the blank TV screen when you first plugged in your HDMI DVD player.

So it was natural that as the CCTV market moved from SD to HD manufacturers would look for a standard – hence HD-SDI being adopted. However, it was a case of Hobson’s choice, and as the standard was designed for broadcast it was inherently unsuitable for the CCTV market, primarily because of the distance required for transmission, but also the cost of suitable cabling and cost of implementation. And the move to 1080p/60Hz and beyond makes the situation even worse.

So analog HD was proposed, but without a suitable standards body such as the IEEE or SMPTE we have seen a fracturing of the market into 3 main camps (I won’t mention our own version at this point). Each are selling well in Asia (except ours!). Each are similar but incompatible. And each, while overcoming the distance/low cost cable problem, have introduced other problems, such as noise, lower resolution (far from HD-SDI resolution) and inability (or at least difficulty to extend beyond 720p/60Hz or 1080p/30Hz operation.

The old analog SD standards were designed as a complete closed system, from the limitations of the eye to colour resolution through to the gamma of the CRT tube. Everything was matched to give optimum system resolution. This changed when flat screen TVs were introduced because they didn’t have the persistence/gamma of the CRT tubes. So quite complex processing (scaling/de-interlacing) was required to match the apparent quality of a CRT tube.

The current CCTV system from image sensor to recorder is mismatched. For example the ISP IC matches the sensor format to the format for transmission – either HD-SDI or analog HD. The MPEG encoder accepts this format 4:2:2 format, discarding some information even before compression (i.e. 4:2:0 format). So we are transmitting information that gets thrown away.

So we therefore have proposed a system that looked at the complete requirements of the CCTV system, which are relatively simple, to record/display a sufficiently high quality image to identify/detect security breaches. Ideally this system should offer uncompressed images as the quality of compression ICs is usually quite poor (compare the quality of an own recorded movie with that of a pre-recorded movie. How can a $10 MPEG encoder IC match the performance of a $25k professional encoder). And as cameras are used many more times than recorders we should try to make the camera as simple as possible (lower cost, less heat, more reliable etc.). This is the philosophy behind SMWE.

I am an engineer, not a marketing person. I offer SMWE as an alternative to the analog HD ‘standards’ but also HD-SDI, because it is designed as a complete system, not one element of it. But of course we need customers to adopt it, we do not make cameras or recorders. But I naively hope that if we offer a better solution then it could be adopted.

What I have found surprising (and please remember I am a video engineer, not a security market expert) is that, from reading your website, the US seems to adopt standards that come out of Asia instead of going its own way. The fact you measure the equipment already makes you a step above most of Asia (try and find a spec. for analog HD cameras). Is there not a US manufacturer that, while obviously being wary of price, actually puts performance first. That can use a standard that is ‘open’ and can be bought from a company that is not effectively a competitor (which is a complaint of many Asia manufacturers who use Dahua’s HD-CVI).

What use is a cheap security system where the image quality is too poor to identify the intruder? (As we are based in Thailand you may have seen the CCTV images from the recent bombing – the only thing you can tell is he was wearing a yellow shirt.)

That is the thinking behind SMWE.

The may get there, in terms of numbers, but not performance. To quote 1080p is largely meaningless. The SMPTE specification relating to HD-SDI, for example 1080p/30 is for a 30MHz luma bandwidth. I believe that HD-CVI is less than 12MHz. So to achieve 1080p/60 they may be able to transmit the format but not at the resolution the related specification requires (60MHz).

Why do you not see this on your images? Because your sensor does not produce 30MHz of luma. A 1080p/30Hz sensor is 1920x1080 pixels. But only half of these pixels are green (which essentially is what constitutes the luma signal). So a 1080p/30 sensor only has 960 horizontal pixels (the rest are interpolated in the ISP).

I don't want to give too much away about how SMWE works - when we did this for aCVi (HD-CVI) Chinese manufacturers took that information and used it make their own designs. But basically, rather than interpolate at the camera, SMWE interpolates just before the display.

So to answer your question, SMWE is as good as the sensor. A poor ISP will actually create artifacts in interpolation, for SMWE the transmitted and recorded image is artifact free, it is a digital copy of the RAW sensor data. This means that you can use non-real time processing to extract the image details which would otherwise be masked in the ISP/transmission/recording.

Just a comment about price. For any new method to be cost effective someone has to design an IC. I can say that SMWE is all digital so it avoids the complication of complex analogue circuitry sharing silicon with high speed digital. (It is likely that this analogue stage. for the analog HD receiver, is the major difference between the performance of the 3 analog HD formats.)

So we are talking with some ASIC suppliers in Korea, China and Taiwan, but of course why would they go to expense of making an IC for a format no one is using. It is a chicken and egg situation.

(Dahua make their own ICs for HD-CVI, but are subsidised by the China government for doing so.)

That said, the discrete solution (non custom IC) can be made affordable as a system because we do not need the ISP in the camera and we can half the size of the HDD (or give you twice the recording time) for the same quality.

We as a company, do not have the resources to make our own IC, else we would of course.