Today, two more patent applications discussing quad-pixel auto-focus sensors have been published by the Japan Patent Office.

The first patent application 2023-167634 deals with the fact that scan times will vary between row and column scanning, which could lead to inaccurate auto-focus performance.

Many conventional image sensors give priority to reading out signals in either the row direction or the column direction, and then move on to reading out the next row or column. For example, when reading out row by row and sequentially shifting the rows to be read out, the pixels in each row are read out at the same timing, but the readout timing for different rows becomes more different as the rows are separated. growing. For this reason, a pixel in which a plurality of photoelectric conversion units for one microlens is arranged in the row direction.

When pixels arranged in columns coexist, the following problems arise. In other words, in the row direction (main scanning direction), where signals can be read out from multiple photoelectric conversion units at close timings, and in the column direction (sub-scanning direction), where signals can be read out at different timings, In this case, focus detection accuracy decreases.

The present invention has been made in view of the above-mentioned problems, and is a main scanning focus detection method using an image plane phase difference method that utilizes a phase difference signal output from an image sensor in which pixels are arranged in a matrix.

The purpose is to make the performance of focus detection similar in the direction and the sub-scanning direction

In this next patent application 2023-167258 with the signal-to-noise ratio which can be different in the row and column directions.

The present invention has been made in view of the above problems, and when the S/N ratio of the focus detection signal is different in the row direction and the column direction in an image sensor in which pixels are arranged in rows and columns , The purpose is to make the performance of focus detection similar in the row direction and column direction

Every now and then I get some patents that I struggle to wrap my head around. Here's exhibit A. Now, first of all, all these problems only happen with a rolling shutter quad pixel autofocus sensor, so if Canon got their act together with a global shutter-based sensor, then they wouldn't be worrying about this ;)

That being said, things like having excess charge flow out of one sub-pixel and into another so they can use that to determine focus more accurately, is something I never thought possible. Perhaps that's where my brain sort of fried when I was writing this article.

As with the article earlier this week here, Canon is delving deep into what can possibly go wrong with a quad-pixel AF sensor (even though it should be noted that at least one of the patent applications mentioned an alternating horizontal and vertical dual-pixel arrangement as well). There's one other time that Canon put a lot of effort and making sure that it was bulletproof prior to implementation and that was IBIS. This is one of the reasons why when Canon's IBIS was released, it was essentially class-leading.

Where there is smoke, there's usually fire, and we're not seeing an uptick in quad-pixel AF sensor applications. Most notably they are describing larger full-frame sensors and finding ways around the problems that these large sensors inherently have.

While we haven't gotten word that the R1 will have a quad-pixel AF sensor, if I were a betting man, I'd certainly lay a few dollars down on that bet. Now the proviso, as with all patent applications, this is simply a look into what Canon is researching, the patent applications may never make it to patents and may not end up into actual products in the future.

Source: Japan Patent Office 2023-167258 and 2023-167634.

Some of our articles may include affiliate links. If you purchase through these links, we may earn an affiliate commission at no extra cost to you.

Go to discussion...

Share.

14 comments

  1. Interesting.
    The OM-1 has a stacked sensor.
    It is possible QPAF it would only work with a very fast read-out.
    I do have to note that the R10 and R7 both have a very slow read-out but better AF than the OM-1.
    It would be great if it would for Canon to only have fast readout going forward.
    I can't see all cameras.
    Maybe for every camera above the R50 is too much to ask for.
  2. The short passage of the first patent seems to describe a new readout sequence in which each pixel in a row is scanned sequentially (like typical sensors) but this occurs in every column simultaneously. Could this be correct?

    If so, it wouldn't be global shutter but it could increase sensor readout speed a thousand fold.
  3. Interesting.
    The OM-1 has a stacked sensor.
    It is possible QPAF it would only work with a very fast read-out.
    I do have to note that the R10 and R7 both have a very slow read-out but better AF than the OM-1.
    It would be great if it would for Canon to only have fast readout going forward.
    I can't see all cameras.
    Maybe for every camera above the R50 is too much to ask for.
    I can't imagine it being in R10ii or R8ii
  4. I can't imagine it being in R10ii or R8ii
    Interesting, I put the chances of getting them at 100%.
    The R8 is pretty much the RP II with a different name.
    I don't see another name change coming.
  5. The short passage of the first patent seems to describe a new readout sequence in which each pixel in a row is scanned sequentially (like typical sensors) but this occurs in every column simultaneously. Could this be correct?

    If so, it wouldn't be global shutter but it could increase sensor readout speed a thousand fold.
    That would seem like a global shutter since there would not be any skew.
    That would create less image noise than a global shutter but more than a typical stacked sensor.
    That sounds like a great compromise.
    Sony is Doomed!
  6. Interesting, I put the chances of getting them at 100%.
    The R8 is pretty much the RP II with a different name.
    I don't see another name change coming.
    That is true, the R8 is about what RPii seemed likely to be. I could be wrong, but I think they would give it with ibis for the mk ii bodies. Why do you think they will have it for lower priced?
  7. Why do you think they will have it for lower priced?
    That is how Canon rolls.
    They sell a lot of lower-priced cameras.
    It is what brought me into Canon in the first place.
    Canon was the first DILC that fit into my budget.
  8. That is true, the R8 is about what RPii seemed likely to be. I could be wrong, but I think they would give it with ibis for the mk ii bodies. Why do you think they will have it for lower priced?
    I’m not sure IBIS would fit in the R8/RP body without making it a centimeter or so thicker.
  9. I’m not sure IBIS would fit in the R8/RP body without making it a centimeter or so thicker.
    Canon has a patent to fit IBIS into a really small body.
    I am more concerned about whether they can power it with such a small battery for very long.
  10. Canon has a patent to fit IBIS into a really small body.
    I am more concerned about whether they can power it with such a small battery for very long.
    Could they change to a larger size battery or do you think it's too unlikely?
  11. Both the IBIS and the battery would be extra weight.
    I would not rule it out but that would be a trade-off.
    It'll be interesting to see how they move forward with the lower priced bodies
  12. I’m not sure IBIS would fit in the R8/RP body without making it a centimeter or so thicker.
    I don’t know how thick the mechanics for it are with Canon, but for the much smaller sensor, Apple and I assume others, have gotten 3D IBIS into their phones. How does the thickness scale up? That’s the question.

Leave a comment

Please log in to your forum account to comment