The new image is optimized for tracking with less severe auto light adjustment. The previous camera had large increments in light enhancement which made object tracking and learning challenging. The new camera is designed for image color/saturation/brightness consistency.

Either your environment is too dark, or the lense has the plastic cover sticker still on it?

We find the image to be clearer with the new camera. Yiu must have the protective sticker on the lense.

LOL. I was just going to ask about the protective sticker, and was searching for the thread I had seen to reference when this came up in the search from DJ posting.

Alan

@DJ Give me some credit... LOL... Of course I removed the plastic sticker(s) and I compared them in the same lighting environment.... Thanks for the explanation. It does track perfectly well, but the image is not nearly as crisp as the gen 1 V4 camera's that I have... I am not overly concerned, I was just wondering....

you will also find less pixelated image with the new camera. The new camera is configured to smooth the image, which means less artifacts and may be what you're interpreting as blurry. I think the "sharpness" of the old camera was misinterpreted by your eyes as being "Crisp" when it was actually pixelated. The pixels of the old camera were large and chunky, due to compression algorithm that wasn't good for motion. The new camera accepts motion much much better with less artifacts.

Also, you would notice a blank white bar appear on the bottom of the old camera, which would also cause a flicker. The new camera motion is smoother with no flicker. The blank bar has been removed as well.

The reason for the blank bar on the old camera was due to light over saturating the sensor because it would auto adjust. The new camera has smarter auto adjustments.

Also, you will notice the new camera frame file size will change - increasing and decreasing with use, which also will result in a crisper image over time because the new camera "learns" the environment. This is because there is a new camera algorithm which will adjust compression to maintain framerate.

And finally, the viewing angle is much greater with the new camera. There's a much wider lens, which provides nearly 30% more viewing area! That's significant.

Generally, you will find the new camera is a much better camera for robot application. This new camera is not a "new camera", per say. It's a new replacement for the previous camera. We still have a higher resolution, wifi enabled, etc, etc, camera coming out 2016 with all of the other items that have been on the pre-order plus more.

Now that we have an experienced SCM (supply chain manager) at ez-robot, product development and manufacturing process has smoothed out and you will start seeing new product often in 2016.

is the new cam in store,does it looks like the older one

You bet nomad, the new camera looks like the old one but the electronics are different. It is a much sexier camera PCB! It has a heat sink also It's an awesome circuit.

so is camera also in the head off jd with you can by in store here

Yes.

@Nomad.... the new (version 2) ezb4 camera is in everything now.... The old (Version 1) ezb4 camera has been discontinued...

@DJ Thanks again for the explanation....

dj thank you&rr

I also purchased the new cameras and tried them for a short moment last night. I also find the picture to be very dark compared to the old one. But my old cam is dead so i really can not compare them. But with the old one I could do colortracking when i had my ceiling LED spotlights on. The new one gets strange colours. Like blue gets green, yellow is grey and so on. It almost look lite when you use a cheap camera without IR-blocking filter.

I read what DJ said about the camera adapting, maybe it will if I leave it on for a while. I´ll try after work today.

Also what does the colours of the LED on mean ?

I think what's happening here is misunderstanding what your eyes see vs what the computer sees. Generally, cameras that are pleasant to our eyes emphasize red, to be warmer. Where a neutral color balance would appear duller. The reason the old camera produced challenges defining colors for tracking was due to this scenario. The old camera had difficulties isolating and identifying reds without a lot of false positives. As you recall, red was detected everywhere on the old camera... In your skin. In anything brown. In reflections. Everywhere! It was a pain to track red - yet, it shouldn't be.

The same applied to blue and green except worse. Green was the hardest to track on the old camera because it wasn't a strong color due to the warmth that made your eyes like it. Green was practically impossible to get right, before.

The new camera, even though it's not totally new - it's just different in the sense that it's optimized for natural colors and tracking. Remember, these aren't go-pros, this is computer vision. Any enhancements or changes made to the platform are to benefit robotics, not home movies or photos of your kids. The contrast and brightness stability makes tracking consistent by limiting level saturation and auto correction.

The challenge with past customer support issues regarding the old camera was poor quality lighting. People would have issues due to their poor lighting conditions, mostly too dark or too direct. Of course the old camera would attempt to compensate for those poor lighting conditions - and therefore return over saturated colors that may look great to your eyes, are not real colors and therefore not great for tracking. These poor quality lighting scenarios would negatively affect vision tracking experienced. People would say "but the image looks fine with my eyes and the computer sees everything as red". Well, yeah, that's because the image levels were modified for your eyes, which affected also every variable for the tracking configuration... Leaving the tracking configuration useless.

The solution to was to make a decision. Are we a GoPro company? Or a robot company?

So, while the newer camera design does have many features enhanced to robotics, it's not a GoPro.

The much wider field lense is also something that will affect the image and lighting. The larger imaged field of view means less pixel detail. So think of it this way, there's far more image but the same number of pixels. Does that make sense? Your robot can see more, but at the same resolution. Increased field of view makes tracking a whole new experience.

You're going to love the new field of view. And I don't mean "oh that's neat" kind of love. It's going to provide a whole new amazing experience. The robot can see further in each direction, which means it can find and track the object easier! A lot easier. It's really really great.

So, long as there's an understanding that ezrobot cameras are designed for vision tracking and image level stability, and that we're not a GoPro - it makes sense.

If you're having issues with past tracking configurations, the solution is to update the configuration settings for the new camera and enjoy the image stability that you are about to experience.

If you find the image is too dark or colors are strange, then the solution is to fix the poor lighting condition.

I would have to say that we have gained a great deal of experience over the last year. By evaluating customer support inquiries and forum issues when the "custom color tracking" tutorial was launched. The number of people with problems tracking colors was high. So we looked at the algorithms first, and they all were flawless in ezrobot's consistent lighting condition. Then we tried in different "living room" and "bedroom" lighting conditions and suddenly we experiences the issues.

Now, we had always been aware of the red saturation condition that returned false positives. I mean, this has been a problem that plagued vision tracking forever - because historically usb web cameras are used. And what are usb web cameras designed for? Making us look good on skype! And that means enhancing colors, specifically reds.

So, now that ezrobot, unlike all other robot companies, has an extremely large dataset of users to analyze from. All our decisions for new product features of changes are based on your experiences. It would be expensive to task ezrobot staff at "fixing" something which wasn't broken. Or purposely making something worse - we would have been out of business long ago

The changes to the camera were performed due to customer feedback of inconsistent tracking experiences and our tests proved it.

The new camera may not make your GoPro friends jealous - but your robot's tracking experience will.

Hi Dj,

Will the new wireless camera coming in the future have this enhancement ? I assume much more will be able to be done due to a wider field of view and tracking ability.

Ron R

Android, the camera available today has the enhancements mentioned in my above post. The camera that is available today is a newer version, even though it was a silent release.

Ok so now i tried tracking. It works way better with the new camera even if the Picture looks a bit odd to me. It was when I used it like a telepresence robot I noticed the different Picture.

What does the different colors on the camera LED indicate? Is it for future use?

Yes, it does track way better.... Green, yellow and blue now track well too... I haven't tried other tracking types yet, but I will later....

@glycol79 you are correct the different LED colors are for future use but also self-diagnosis. Something we hadn't really mentioned yet is that this new camera is also smarter

If I have time I will compare the new camera vs old camera on video for everyone!

Awesome - i knew you'd love it and hope my lengthy explanation helped

The RGB LED flashes for different "boot up" sequences - if there is a custom support call regarding an inoperative camera, we can diagnose the issue by the color. I don't have the color codes in front of me at the moment.

Also, there's something really interesting about this new camera pcb as well. If the camera sensor gets damaged, the software on the chip will recognize it. And to tell the user that the sensor is damaged, the software will display a message in the ARC camera window.

How is the new camera at face tracking, the cameras I have now see a lot false positives. Also, is the new camera better at seeing QR and Gylph codes at a distance? I have found that they have to be within a few feet of the camera to be read.

It's all better except face. That's still an ugly algorithm. Some upcoming plugins will be blowing your mind shortly though. Stay tuned!

I don't hear anything about motion tracking. Did I miss this? Color tracking is good if you wear a red or green shirt but what if you want your robot to track someone who is walking past and move his head to follow. I had little luck with this on the old V3 camera and gave up. Will the new V4 upgraded camera be able to do this better than even the old V4 (which I haven't even powered up yet BTW).

Motion tracking will move toward the direction of a changed object. It does not require a special camera, as it's a very primitive algorithm in which two frames are compared against each other for differences. You can find more about what tracking types are available and descriptions of each tracking type and what it does by pressing the ? (question mark) next to the X (close) on the camera control - or hovering your mouse over the ? (question mark) in the camera control by the types.

There is no tracking type for what you described.

Also, the camera itself does not have any "brains" for tracking. Doing so would be limiting to the functionality and expand-ability of the camera - which is why all embedded camera tracking products have pretty much died off. Offloading processing capabilities to the PC is the only way to provide real tracking due to the higher processing power of a PC vs microcontroller.

The discussion in this thread is regarding the camera image and the brightness/color levels for tracking types. This thread does not discuss any enhancements to the tracking features.

OK, thanks for the info.

Anytime I always write overly descriptive responses so others who don't know what the question is about learn as well