CMOS comparison AR0521, OV4689, IMX291

Guys is it me or does this new Aptina AR0521 look better in every situation to me. Any good dash cams using it. I am tired of these 1080p Sony sensors.
Maybe is looking better only during day. People are paying much money on dashcams because they want good image also during night.
Because you are tired about Sony 1080p sensors maybe is time to prepare for the Sony 2K sensors like IMX335 Starvis.
Right now a good dashcam during night can be only with a Starvis sensor from Sony. Maybe in next 5-10 years Aptina will create the same which is creating Sony now. ;)

enjoy,
Mtz
 
Maybe is looking better only during day. People are paying much money on dashcams because they want good image also during night.
Because you are tired about Sony 1080p sensors maybe is time to prepare for the Sony 2K sensors like IMX335 Starvis.
Right now a good dashcam during night can be only with a Starvis sensor from Sony. Maybe in next 5-10 years Aptina will create the same which is creating Sony now. ;)

enjoy,
Mtz

Yeah I made that comment before I did hours of research. So from what I gather many brands look good in the day but the SONY IMX291 is the best at night since it uses large cells that allow more light in but using larger cells limite resolution. Plus people have said the Sony has better motion resolution due to it's shutter speed.

Isn't the new 1440p Aptina AR0521 sensor close enough to the Sony at night after the custom Firmware updates from DOC
 
Hi. The starvis uses special low light technologies and not just bigger pixels. I'm still waiting for 4k starvis, should still be nice low light.

It could take 17 years for all these technologies to be available to OnSemi fur Aptina sensors, or if third party until license finishes or patent expires. There are other low light technologies which they could pickup.
 
aside from pixel size Starvis are BSI sensors which works better than an equivalent pixel size FSI sensor, 4K sensors will still need to be a bigger format to do as well, smaller pixel size would still put them at a disadvantage compared to the larger pixel size of 2MP sensors, there's more to it than just the sensor used, there are cameras around that use good sensors but still don't do that well in low light
 
I read about a phone with a 1/2 inch 48 megapixel main camera ( IMX586 ) , when low light you can do a 4 in 1 pixel binning to get better low light performance and so give you "a pixel" of 1.6 X 1.6 micron instead of the actual 0.8 X 0.8 micron each pixel are.
But to me it feel like with these high pixel count phones, well shooting pictures in the binned mode and so a lesser resolution are often better than if you shoot pictures in the sensors native resolution.
Which are strange to me as i am talking about daylight pictures here and not just night time pictures.
Also taking those low light pictures well it seem like to be several pictures stacked on top of each other, and so take a while to snap a picture like that in low light and so probably wont do well with movement in the frame.
 
I read about a phone with a 1/2 inch 48 megapixel main camera ( IMX586 ) , when low light you can do a 4 in 1 pixel binning to get better low light performance and so give you "a pixel" of 1.6 X 1.6 micron instead of the actual 0.8 X 0.8 micron each pixel are.
But to me it feel like with these high pixel count phones, well shooting pictures in the binned mode and so a lesser resolution are often better than if you shoot pictures in the sensors native resolution.
Which are strange to me as i am talking about daylight pictures here and not just night time pictures.
Also taking those low light pictures well it seem like to be several pictures stacked on top of each other, and so take a while to snap a picture like that in low light and so probably wont do well with movement in the frame.
If you take a modern plastic framed window with a single sheet of glass and replace it with four separate pieces of glass joined by strips of wood, the window is still the same area, but you get less light because the wood blocks a lot of light. That is the equivalent of going from 1080 to 4K, to reach your 48 megapixel sensor you need not 4 pieces of glass, but 16 pieces, and the wood will block so much light you will have to use expensive lead to join the pieces, but you will still loose a lot of light!

4K sensors will never be as sensitive as 1080 sensors, but sensors are getting more sensitive anyway so a next generation 4K sensor may do as well as current generation 1080 sensors, which are good enough for many people. Then again, having seen the next generation 1080 sensors, I think I might still stick with 1080!
 
aside from pixel size Starvis are BSI sensors which works better than an equivalent pixel size FSI sensor, 4K sensors will still need to be a bigger format to do as well, smaller pixel size would still put them at a disadvantage compared to the larger pixel size of 2MP sensors, there's more to it than just the sensor used, there are cameras around that use good sensors but still don't do that well in low light

Yes, I'm saying that starvis does so well at fullhd, its going be OK for me at 4k. As far as recording preincidents coming up the road, 4k is better.
 
I follow a bit about camera design. The interspace area between the pixels (wood) often does not loose much, and there are ways to use the interspace to collect light too (which the company fillfactory used to do).

The amount of pixel to compared to the space its in is called the fill factor. I can be high like the 90% foveon x3 sensors used in Sigma DSLR's, which aren't know for good low light, to below 50% on sensors. However, will probably find the difference between the fill factor on a larger pixel on bsi and those smaller pixels, not to be much.

As far as I'm aware, that Sony doesn't have starvis, so I would not expect much from it anyway. If there was a 0.8 micron starvis sensor it might stretch it a bit. Color pictures by starlight is very low luminance, so, there is room there before it becomes worse than what our eyes color vision can do.

Still, there are 4k starvis sensors (often used I'm security cameras). It would be good to get a comparison to one of those? I was looking it up recently. Maybe Blackview car cameras has one?
 
Please watch to the end at this great side by side comparation and observe that crystal clear image at night from that NON 4k Sony sensor, IMX 123, 3mpx.
Way better than the 4k Sony IMX 226, 12 mpx.
I think this IMX 123 represents the future of dashcams.
 
My poor eyes, that was a dreadful comparison video. Its too late at night here for that. The 4k starvis actually showed more in some shots because it was boosted. You would actually be able to see that attacker crouched beside the car in the carpark.

It's not a very good comparison comparing a car cam to a security camera of that type boosted and not even framing up the shots the same.

You are going get more evidence collection during the day at 4k, and as I said, the 4k is good enough for me at night. What do you want to film with the lights off anyway, you doing night riding Dadi? You need more than starvis for that, things can't see you to avoid you.
 

4kp60 12 bits, 1/1.7 inch nice.

Just did a pixel area comparison between the two sensors, because the 4k is so big, it's pixel area is over half of the other one, so leas than double the light gathered or one high bit worth. Not going make much of a difference in the car lights. Maybe you could see double down the road something. But it is a newer chip, so it might work better and recover some of that, but I have no figures to compare.

That document is nice, you can see that it is picking up a much wider viewing angle, allowing a much faster lens to be used to get more light.
 
Last edited:
Back
Top