Friday (yay) and time for another look behind the scenes of Memoto’s production. If you have missed out on why we’re doing this on such a regular basis, check back on previous updates and most specifically this one.
Here’s what’s been accomplished this week:
- Our industrial designer has made CMF spec (Color Material Finish) to hand over to our manufacturers. Now they will now how every surface and color will look and feel on the hardware, which obviously is a crucial input in their tooling for the camera case.
- Blueprints are finished for the USB cable (micro usb to regular usb) that will come with the Memoto camera. Let’s leave the details on the looks of it for you to find out when you open your package, but this will be one gorgeous little USB cable. (NB: gorgeous, but not slowing down or risking the rest of the shipment.) Next step is to start the tooling for the cable.
- On the packaging side, we’re waiting for new samples of the actual package, the packaging sleeve and the instruction manual. The ones we got so far looks good, but they needed some polish.
- We’re waiting for the plastic tooling to inject and mold the various parts in the Memoto camera. It will take a few weeks more, and we hope that no issues creep up during that time of course. The injection molding tooling in this case is quite advanced for producing the tiny plastic shell with its many features, so when we are done we look forward to doing a quick video showing just what machinery is involved.
- We are also still waiting for the first 75 PCBs to get ready and are preparing the surface-mount assembly of them right now. We have high expectations for the assembly quality as we collaborate with a very good manufacturer, but this is one area where yield issues can creep up so we have to put these assembled PCBs through mechanical and environmental stress tests before ordering the other thousands of boards.
- The production test development and setup is also ongoing, with PCs being set up at the factory now which will inject the production test firmware into the assembled boards and produce a go/no-go signal. We would be happy to share some videos of this process as well when we get the opportunity.
- Work has been made to stabilize and speed up the process of uploading pictures and creating moments. Today, images are lined up in a que directly on our servers as they are being uploaded, thus creating a heavy load, a drop in speed and an overall not-optimal-experience. When we’ve put Amazon S3 in the middle between the uploading client and our servers the capacity for the que of images will be (almost) infinite and keeping our servers out of the action during the upload process. On the “momentification” side, we’re teaching the algorithm to build moments as pictures are uploaded and not in huge batches as uploads are finished. The result: a speedy and robust upload experience.
- To increase the accuracy of the momentification, we’re also starting to collect data from humans. (Who would have thought, huh?) This way we can test for example that the algorithm’s opinion on when a moment begins and ends correlates with what the human user think about his/her memory.
- The OSX client in older versions of OSX has been a pain. We’ve used new API:s and frameworks to build the client, but these haven’t been supported by older versions of OSX. After new custom-written code, it works like a charm.
- For the Windows client, the last steps of the design process and the first steps of coding has been initiated. Iterating through a number of design alternatives for the OSX client and tested them on people (humans, again!) will prove useful here.
Anything you think is unclear or that you want to hear more about? Let us know in the comments!