Video Enhancement Startup (Lentix) Partners With Milestone

Huh. The cameras in their video are some of Avigilon's demo site cameras.

Reactions:

(2)

John Honovich

In reply to Undisclosed Integrator #1

IPVM

I asked them about that. They are not integrated into Avigilon's VMS but they said that Avigilon let them use some of their streams for testing purpose.

Reactions:

Undisclosed #3

In reply to John Honovich

IPVMU Certified

So the demos are not of live video?

Reactions:

John Honovich

IPVM

I am not sure if that makes a difference. To this process, it is just ingested video, whether its live surveillance feeds or your favorite Miley Cyrus music video.

Reactions:

Undisclosed #3

In reply to John Honovich

IPVMU Certified

I am not sure if that makes a difference...

I think it makes a big difference and here's why:

It's supposed to be instantaneous realtime enhancement right?

Watching that video certainly reinforces that claim.

So, if one purchased it and therefore expected to see real-time enhanced live video but that enhanced video was delayed 4 seconds, one might be disappointed, correct?

Yet in the video they seem to be virtually in-sync, frame-for-frame, surely a wonderous feat of image processing. All the more amazing that such a computationally heavy process (as they claim) never introduces even a single frame offset on those multi-mega-pixel Avigilon clips!

Ok, it could be, what do I know?

But since the demos are not 'live' there is no way to be sure that the clips are not re-aligned after the fact, ostensibly so that one can easily see the enhancement delta. If they were adjusted, then it should be stated.

To this process, it is just ingested video, whether its live surveillance feeds or your favorite Miley Cyrus music video.

No, when it comes to image processing these are not the same at all. Unlike unbuffered live video, the Miley Cyrus video offers what's known as 'look ahead' processing, where the process makes decisions based on things ahead in the timeline. This knowledge reduces artifacts and a greater degree of enhancement.

Of course if you want a little 'look ahead' processing on live video, you can always buffer as much as you want, but there goes the instantaneous enhancement.

Take a look at this other, unnarrated Xpose demo video.

Notice how the the moving car is ahead by > 3 secs in the opening shot. Also, notice how the latency seems to change at times to almost nothing. This is possibly a more realistic example of what you might get if you purchased Xpose.

Final observation, I don't see a whole lot of movement in any of these clips, maybe the processing requirements are more than anything dependent on scene complexity / movement.

Therefore I feel yes, it is critical to see a live enhanced stream, side by side with the unenhanced, preferably with various complexity and motion.

I will download the demo when I get back to the lab. ;)

Reactions:

(1)

John Honovich

IPVM

That was a very long way to say, "What about latency?"

Latency is definitely a concern, I don't know if it's critical though. Of course, it depends how much latency - 200ms, 2s, 20s, etc. But, I would think even 2s is probably not a big deal for most applications unless you are providing commands to a responder or trying to control a PTZ at the same time.

The bigger issue for most users is just how much better the video becomes.

Reactions:

Undisclosed #3

In reply to John Honovich

IPVMU Certified

The bigger issue for most users is just how much better the video becomes.

That can be the biggest issue, that's ok with me.

Anyway, after reviewing the videos they are clearly saying that the enhancement is taking place just as it is shown, without a perceptible delay.

So pretty amazing.

Reactions:

Undisclosed #3

IPVMU Certified

So pretty amazing...

<edit> So pretty amazing the balls these guys have...<\edit>

Reactions:

(1)

Undisclosed Integrator #4

The demo video is obviously doctored. The enhanced video on the right is a few seconds ahead of the original on the left. Now thats real enhancement showing something a few seconds ahead of realtime. That is amazing.....This is not latency but rather its predicting the future by 3 seconds ....what an amazing break through in surveilance.

Reactions:

John Honovich

In reply to Undisclosed Integrator #4

IPVM

As much as I like to criticize manufacturer marketing material, I am not sure the point of putting so much emphasis on their demo videos.

They have a free trial. If you want to check it out, check it out.

We may try it out tomorrow in between Axis analytic and another camera test.

Reactions:

Undisclosed Manufacturer #2

...but does it really work...?

Reactions:

Undisclosed #5

Curious why they use Avigilon cameras, fact is that if the dynamic range is not captured then it cannot be shown! Looks that they do some video processing as well but if the dynamic range is not there, there is not much the tool can do without raw material.

Reactions:

(1)

(2)

(1)

Undisclosed #3

In reply to Undisclosed #5

IPVMU Certified

...fact is that if the dynamic range is not captured then it cannot be shown!

But is what is shown all that has been captured?

Reactions:

Undisclosed End User #6

For monitoring purposes, I could see this work. However, for evidential purposes, I would not think this would be acceptable due to the fact this is software and not a filter changing the data. Once software changes data, then the picture is not 100% official even though it looks more like the real thing. The-hair-above-average person does this with simple photo software to the more advanced Photoshop software when dealing with personal photos. In these programs, it is known as an "Effect". It is not live video software, however, I'm just pointing out it is a software effect. When it comes to software editing of images, if we allow this, then we allow the possibility of “Photoshoped” evidence in our judicial system.

This is something you should ask the company and others what they think is acceptable software editing when it comes to evidence. Where should the line be drawn?

Reactions:

(1)

John Honovich

In reply to Undisclosed End User #6

IPVM

Good point.

They mentioned they keep copies of the 'raw' video and the 'corrected' video. If it ever got to trial and a suspect contested it, you'd probably need to go to some expense to have an expert analyze and verify the 'correction'.

Reactions:

(1)

Luis Carmona

In reply to Undisclosed End User #6

Geutebruck USA • IPVMU Certified

I'd disagree. All pictures are computer digitized images. Camera with Wide Dynamic Range have modified the final pictures to a certain extent. I'd think the only time it would really be a contention if is was a single video clip altered by a person after the fact. If many or all streams are being enhanced [altered] automatically without thought influencing a future outcome, I don't see it as being that much an issue.

Reactions:

(1)

John Honovich

IPVM

In reply to Luis Carmona

Ultimately, it's for a court to decide, if someone objects. What someone could argue is that the automatic alternation skews or misrepresents reality (e.g., the software makes it seem like a person is holding a gun when in fact it's a shadow).

Reactions:

(2)

Ethan Ace