Serial Camera, Courtesy Of The STM32F4

Look around for a small, embedded camera module, and you’ll find your options are rather limited. You have the serial JPEG cameras, but they’re rather expensive and only have VGA resolution. A Raspi, webcam, and power supply is a false economy. GoPros are great, but you’re still looking at some Benjamins used.

The guys at GHI Electronics are taking a different tack. They’re using image sensors you would normally find in cellphones and webcams, adding a powerful ARM processor, and are still able to sell it for about $50. It’s called the ALCAM, and they’ve stumbled upon a need that hasn’t been met by any manufacturer until now.

On board the ALCAM is an OV3640 3-Megapixel image sensor. On the back of the board is a STM32F4 and a microSD card slot. The board can be set up for time-lapse videos, stop motion animation, or all the usual serial board camera functions, including getting images over a serial connection.

The ALCAM operates either connected to a PC though a 3.3V serial adapter cable, through a standalone mode with pins connected to a button or sensor, to the SPI bus on a microcontroller, or a serial to Bluetooth or WiFi bridge. Images can be saved to the uSD card, or sent down the serial stream.

It’s a pretty cool board, and if you’re thinking it looks familiar, you’re right: there’s a similar DSI camera/STM32F4 board that was an entry to The Hackaday Prize. Either way, just what we need to get better cameras cheaper into projects.

47 thoughts on “Serial Camera, Courtesy Of The STM32F4

  1. This isn’t a hack. It’s two $3 parts, $1.50 worth of connectors on $3 of PCB doing exactly what they’re designed to do. The only link here is to a kickstarter.

    Can we stop with all this constant kickstarted bullshit please?

    Posts like these dilute the essence of what people love about hackaday, so if you can’t find the hack in this, I urge you, please remove this blog entry.

    1. So they glued an STM32F4 to a camera that has onboard compression just for serializing the data? Why?

      That SoC is perfectly capable of driving a cheaper camera and doing its own compression. The bells n whistles STM3240G devkit for this chip even has a 2MP cam onboard with source showing how.


    1. …Just wondering… I wonder if “Still Image Host” in LUFA means it can talk to webcams through USB. All of a sudden, I feel like I might be warming up to this tool.

    1. in the past nothing, but then they said no more kickstarters. But now they’ve been showing them more often (mostly scams being debunked) so I’d still assume nothing.

      1. We receive nothing, so any complaint that this is advertising disguised as news is false. We also never said we wouldn’t be covering Kickstarters/Crowdfunding, so any claim of a change in editorial direction is false.

        Five years ago, you would have seen something like this up on some guy’s blog, and you would have asked if he was ever going to take it to production. Now, you see it on a crowdfunding site and you’re complaining it’s all just an ad. That’s just an observation – I’m not calling specific people out for being inconsistent, merely the group mindset behind the complaints about Kickstarters being inconsistent.

  2. OK Its kickstarter, but a man’s gotta eat.. whoever is trying to produce this has decided to actually commercialise it rather than just do it for the heck of it, which isn’t a crime. Presumably the added value in this product will be the firmware. I agree its not really a hack in the strictest sense, but its still interesting none the less. As to the false economy of using or not using a pi.. I suspect the pi consumes way more power, and is pysically larger but in turn is way more flexible. If you want a camera to fit in a small space and run on batteries, but dont want to risk your GoPro this device makes some sense. If you want to incorporate a camera in a new or existing product, but have limited space, this makes sense. Other use cases may apply, but dont knock it simply because it is a kickstarter, or we are entering vi/emacs arduino/555/discreet transistors territory and no, I have no connection with this product, so no reason to defend it… but if they were to send me a free sample… }:¬). … I’m sure I could find many uses for it.

    1. If you only want a cheap camera for recording (but not need to be able to stream the data which this is supposed to have), you can buy $15 “HD Camcoder” from DX. Use analog mux to isolate the SD memory to your microcontroller to access the recorded data and GPIO to control the camera.

      Doing a project like that would be more a hack than *buying* crap from a kickstarter.

    1. What’s wrong with the price? If You do cheaper, please, kickstart it, I gonna support You. Any electronics device costs more than sum of parts soldered on it. BTW software is the most expensive part in such devices because You need to support and maintain it.

      1. I disagree ! if you do a kickstart project you do it like you do some software under GNU linux: for free in your spare time because you love it ! so the goal is to provide cheaper than what a guy could buy alone. 10 000 cpus are cheaper than 1 and share knowledge.
        Now if you want to make money as any capitalistic bast*** well kickstart shouldnt be where u go.
        thats why you see all those negative comments about kickstart project. At the beginning people there were doing it as a project for a community without making money in the plan. Now there ‘s a bunch of gold digging bit**** on those sites.

      1. :) Ibrahim and I will be launching the OpenMV KS in the next several weeks, details to be announced on the Google group Join and stay tuned.

        OpenMV will be ~$50 with vision processing and Micro Python onboard; uSD, LED, SPI, I2C, Serial, USB. WiFi shield is in the works plus others. Fixed sensor with interchangeable M12 lens for better reliability. Enough backers means an upgrade to faster STM32F427 or ‘429 both with more memory thus capability. OpenMV is, well, fully open software and hardware. And hackable.

        We already sold and shipped 18 hand-built (by me) OpenMV cams for folks willing to test.

  3. These cellphone cameras are seductive. They’re cheap, good and small, perfect for someof our projects… I’d love to use them but OmniVision, in particular, is a difficult company to get information out of. They part with specs only after signing an NDA, and even then, there’s no apps assistance unless you’re Apple or another of their “first tier” customers.

  4. If you’re looking for a source for a *super* cheap VGA resolution camera, Electronics Goldmine has ADCM-2700’s. The image quality isn’t the best (seems to have a short focal length). But for simple machine vision, video chat, etc, these are fine. Datasheets and example source code are hard to come by…but I was able to get them to work without too much fuss. No JPG output, but very flexible scaling options and push data out via 8bit parallel or synchronous serial.

    They’re at

  5. Wish this was not vapourware. Also wish they would use a REAL lens mount instead of the super junky cellphone camera assembly.

    I want a threaded security camera mount or better yet a C mount. that way you can change out to a fisheye or a telephoto if you need to as well as far better optics than the tiny glob of plastic.

    1. Vapourware? GHI has working prototypes and their history and catalog of existing devices should give you plenty of confidence that this will happen. This is not their first Kickstarter. GHI doesn’t typically start a Kickstarter unless they have a product about 99% complete. Regarding lenses, the Kickstarter says they are adding various options for that.

  6. I’ve used this camera before but it had a parallel interface for raw YUV image data plus I2C for setting up a ton of internal registers. (And getting the relevant data out of OmniVision was an odious task.)

    The article mentions this one is using MIPI DSI but how does that work with the STM32? Surely too fast to be bit banged and decoded on on the fly?

  7. I don’t mind seeing the Kickstarter stuff on here. I’m interested in projects like this so I’ll bookmark it and later when I have the need I’ll look it up. Even if I don’t buy the product I now have some information on what devices work together and how to take a general approach. It’s notable that this article has generated some useful comments, also worthy of bookmarking.
    The value for me is the idea. Someone has done the basic R&D and even if someone else is just putting parts together and it’s new to me, it’s useful.

  8. I’ll probably buy it but wait till someone writes energy efficient motion detection code so I don’t have to do it.. I want to use it on cell powered surveillance.. If you weather proof it, it’d be good for agriculture and RC aviation too.. Maybe even make optics for it..

  9. No one has mentioned the even-more-capable CMUCAM5 (aka Pixy). $69 on Amazon, $75 on Adafruit:

    Also for a lot of scientific applications, you need the highest quality signal, meaning you don’t want any compression. So for my purposes unless they added a RAW capture feature, the ALCAM is almost useless to me.

    Also, the raspi camera suffers tons of digital noise in low light… so either someone messed up the board design (maybe I just have an early revision), the compression is set too high, or some combination of the two.

  10. Looks cool, but it is a kickstarter, and I’m so tired of kickstarters, they are so 2013. I mean, it is getting to be that you can’t just buy anything anymore, you have to make a pledge, wait 30 days or whatever, hope it gets funded and then wait another 6 months, by which time you totally forgot why you wanted one in the first place. Then maybe you get it and maybe it is really cool. But I want to just to buy one now. ARGH!

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.