Neuros OSD Review

An anonymous reader writes "The Neuros OSD promises a lot — it claims to be the first open source Linux-based embedded media center and it "records video and links your PC, portables and entertainment center". Bold claims, but can it live up to them? Linuxlookup.com has a two page review of the Neuros OSD."

Comments

I'd get one if....

[networkBoy]

I've really been thinking about getting one of these. I currently use the Xbox, mod'd to run XBMC. If the neuros can work that well out of the box (preferably running a port of XBMC) then I'd be sold. As I see it now it's still too much work compared to the Xbox solution.
-nB

No 16:9?

[riceboy50]

According to TFA, this does not yet support 16:9 aspect ratio. Presumably that also means it doesn't support HD content. I wasn't clear if this is supposed to be a new DVR solution, but if so, then it needs to support HD and 16:9 before I'd consider it.

Re:No 16:9?

[Junta]

Also lists resolutions, which top off at 720x480. Definitely not HD preserving, and likely no HD tuner. Of course, an HD tuner PCI card is about $100, so it may have put the vendor over their price point.

Doesn't take much to support 16:9. Also is effortless to support most sets with HDMI and distinct audio ports with a DVI-HDMI cable.

I have a diskless mythfrontend on my TVs HDMI port, running 1280x720, and the backend has an HD Tuner. Absolutely beautiful.

Not All Three

[nbannerman]

I RTFA and noticed the following snippet;

I have quite a large TV - a 37" widescreen - and the playback on that screen was quite poor. Whether this was related to the widescreen issues mentioned below or the general quality of my input source I'm not sure, but I certainly won't be using the OSD to record TV for playback on my TV.

I'm not sure I can justify spending money on something that'll record stuff that only looks good on the PC or a portable player... if I wanted that I'd just use my PC's inbuilt functionality. Still, a good start.

Re:All I can say is...

[Ken_g6]

I thought you were going on a scale of "Ouch!", "Wow", or "Boinnnnnnnnng!" Somebody's been watching too many old Christmas movies. I guess it's me.

At $230, it doesn't look like it's breaking the bank for a DVR...until you realize that it doesn't include a hard disk! It also doesn't record HD video. At that price, it seems like it should do one or the other.

Review

[dj961]

The Neuros OSD is small. Very small. At only 14cm wide, 14cm deep and 3.2cm high, it fits comfortably in just about any hole you'd care to stow it.

Just lube it.

Goatse?

[scribblej]

The Neuros OSD is small. Very small. At only 14cm wide, 14cm deep and 3.2cm high, it fits comfortably in just about any hole you'd care to stow it.

Oh my!

I'm trying SO hard not to post a link to a certain hole.

Buff it up to 1080p and add cable card support.

[zorkmid]

And then you'll have something. As it is now it's not terribly useful.

First?

[RedWizzard]

first open source Linux-based embedded media center I think not. MythTV based systems have been commercially available for a while.

Re:First?

[|<amikaze]

Please note the word "embedded".

Let's see...

[MBCook]

Widescreen? Nope. HDTV? Nope. Dual tuners? Doesn't look like it. Display on front to show what it's recording? Nope. Support for digital cable (cable card)? Nope. Downloadable programs over the 'net? Nope. Suggestions based on other users TV viewing? Nope. "Season Pass" like recording? Doesn't seem like it. Fits nicely in a rack of home theater equipment (doesn't look like a PC)? Nope. Ability to hack? Yes. Monthly fee? No.

Let's compare that to a TiVo series 3.

Yes, yes, yes, yes, yes, soon, yes, yes, yes, no, yes.

It only won in the last two categories.

Yeah, I'll drop my Series 3 for this thing. Heck, I wouldn't drop a Series 2. You can hack a Series 2 to add other stuff, and still have the great TiVo UI and service.

I've yet to see what I consider to be even a mildly compelling alternative to a TiVo. Unless you have all the parts sitting around and want to build a MythTV box for free, they just aren't there. I mean, why should I choose this over a cable company DVR which would give me things like On Demand and HD?

TiVo: Still #1, no serious competitors since the death of ReplayTV.

It's not Beta.. It's not even FC

[lushmore]

I had an OSD for a couple weeks and returned it. I'm an embedded Linux developer by trade, so it would have been right in my wheelhouse, IF I had a ton of free time to work on it and time to wait for Neuros' and others' contributions. But seriously, you can't call it beta if > 50% of the features on the box don't work reliably. It's not fair to review the unit at this time. It's nowhere near done.

Ask someone who knows...

[gillbates]

Neuros bought most of the video codecs and Linux BSP from a third party. How do I know? - I was an engineer for said third party, and wrote/developed a few of the modules for this platform. In fact, I'm intimately familiar with the video and audio playback code.

That said, I have a few comments:

The sound and video often get very out of sync and sometimes the video judders, or slows to get back to where it should be...[emphasis added]

First of all, I had not observed this at all when using a pristine source. We did recognize that our coping mechanisms would produce a similar result if the incoming source had missing frames or audio, etc...

In fact A/V sync was one of the enduring problems on which I worked during my tenure. Suffice to say, we chose to gradually pull audio and video back in sync when sync was lost because our clients complained that the alternative appeared too jittery. Unlike other vendors, we could present acceptable quality playback with as much as 1/4 of the frames missing from the input stream. Most other encoders/decoders would produce a noticeable, annoying frame-jitter whenever there was a loss of either audio or video.

But, aside from that, here are some more things the article failed to mention:

1. The processor is a TI DM320. The ARM 9 can run at ~200 MHz, and the DSP at ~100.
2. The DSP, not the ARM, does the video encoding/decoding. The DSP or the ARM may be used for audio playback, depending on the codec used.
3. Typical ARM CPU usage during playback is less than 50%. When audio is done on the DSP, you can easily attain less than 10% CPU on the ARM. The DSP does the overwhelming majority of the work on this platform.
4. We ported MicroWindows to this platform. Though it might not be in the BSP, you can definitely run it on the Neuros if you have the time to port it.
5. It has 64MB of SDRAM. The address space of the processor allows for up to 256 MB to be installed, if anyone is interested.
6. Likewise, you can replace the 4MB flash memory with a larger one, if you'd like. Typically, there is about 256MB of address space per chip select.
7. You should be able to get X up and running on this platform. Some of our past clients used GTK on this same platform.
8. If you are interested in debugging, telnet to the board and look at the /proc directory. Several of our modules will list interesting statistics on video and audio performance such as: Number of dropped/delayed/skipped frames, audio/video sync difference, number of frames, number of interrupts, etc...
9. The platform does support IDE drives. Someone willing to get out a soldering iron and tinker could very easily add tivo-like video recording to this device. But why would you do that when...
10. The device is designed to be used with USB hard drives. The reason why an HD isn't included is because the usage model is that of a user plugging in an external USB HD, and encoding to that. That way, you can expand the recording capacity of the device without cracking open the case as above...

Hope this helps.

I do feel some connection to this project because I did a lot of work on this platform. Truth be told, I'm thinking of buying one just for sentimental reasons; unfortunately, my company didn't hand out samples. I do know quite a bit about the BSP, and would be happy to answer any questions regarding the platform that I can.