Tuesday, November 24, 2020

Sony Promotional Videos: SWIR, Polarization, ToF

Sony publishes 4 promotional videos about its image sensors: SWIR, polarization sensing, and ToF:


  1. I'm often surprised that this promo videos miss to mention one point that in my opinion is one core benefit of this Sony polarizer and SWIR sensor - the CMOS ROIC identical/similar to their standard products. They seem to try to find/present applications and miss to mention to the guys that do the application how easy it is to apply. The polarizer/MZR for example is electronically identical to the standard sensor, so no hardware/firmware difference for the camera, you just solder a different part on your pcb (this the main reason why all the industrial camera suppliers that have a imx250 camera also have a mzr camera - it gives you a high margin product more or less for free). Even more the SWIR - the hurdle to jump across to get a decent swir camera integrated in a form factor and price point is significantly lower than before. Most of the complexity is hidden behind a digital interface, a small team can integrate the sensor successfully even if the team is not a camera manufacturer team but a electronics team at a robotics company for example. This is more important in my opinion than perfect image quality that is often the key message in promo videos. I posted the same things various times before... e.g. camera projects in my environment (robotics, industrial automation) did never fail because of image quality being too low. The killer was sometimes when a sensor was apparently hard to apply (i had a example with a swir sensor btw. such projects tend to be stopped in advance because if the risk/effort estimation is given with a lot of 'i dont knows'. If you read the datasheet of the sony swir sensor and already know a different Pregius, a effort estimation guess is low and with confidence and this gets the project started and sensors sold. In my opinion image sensor companies overestimate the importance of image quality and underestimate the importance of applicability of their products. It was sometimes funny to watch in details. For example... the SLVS interface of the imx990 is stunnilgy similar to the sLvds of the IMX250, but you dont know this in the beginning because the docu dont tell you. The first impression is - the interface is a different story. But in the end what you need is an extra resistor - thats it. But this info is hidden in a way that when you do the effort estimation you dont have it. You start going for IP cores and negotiate a few extra 10ks in budgets etc. (which can stop the project befor it even started) before you reach the point where you find out - quite surprised - that in terms of firmware this is the same interface and you run it with more or less the exactly same firmware. Integrating the IMX990 is like "walking across open doors after expecting you have to dig holes across one wall after another". Well... I think companies could win a few projects more when the "how to apply my product if you know xy concepts" would be answered (also the sales guys dont tell you this, they tell you how good the image quality is). Because projects get started or stopped based on this info. Often developers have quite an influence on decisions if a project gets started or not because the bosses take the decision based on this guys infos that were created after a few days of datasheet/doc study. The better the question 'what effort/risk is it' and 'what is to do' can be answered, the better.

  2. In the past 20+ years I've never heard SWIR pronounced as "swire" - only "sweer". That's a first. Hearing it pronounced that way in the Sony video sounds annoying TBH!


All comments are moderated to avoid spam and personal attacks.