How does multi-point focusing work? Does it increase DOF?

Discussion in 'Panasonic Cameras' started by New Daddy, Oct 10, 2011.

  1. New Daddy

    New Daddy Mu-43 Regular

    Jan 24, 2011
    When there are several people in the frame and they are not situated in the same plane (e.g., around a round table), can "multi-point" focusing of Panny solve the problem of getting the focus on everyone's face? Does it increase the DOF by changing the aperture? I did some trials but the "multi-point" focusing doesn't seem to change the aperture. If not, is the only way to get everyone's face in focus changing the aperture yourself to increase the DOF?
  2. Howi

    Howi Mu-43 Veteran

    Feb 23, 2011
    Real Name:
    Point 1 - how can a camera know what you want in focus?
    Point 2 - Focus points and aperture are NOT related
    point 3 - Your the photographer, YOU need to decide what you want in focus and use the appropriate aperture to get the desired DOF.
    point 4 - take control of the camera, don't let it control you.
    point 5 - If cameras could do everything, why would they need anyone to operate them?
    point 6 - Photograph is a science and an art form, the camera is only a means to an end, it's the photographer that decides who, which, what, where, NOT the other way round.
    point 7 - The effort you put in is rewarded by the results you acheive,
    point 8 - If you are not getting the results you want, refer back to point 7.
  3. hkpzee

    hkpzee Mu-43 All-Pro

    Sep 5, 2011
    Hong Kong
    Real Name:
    I think New Daddy has asked a totally valid question, so there is really no need for condescending comments.

    Unfortunately, I don't have a Panasonic camera, so I don't know how that new "multi-point" focusing feature works, or how well it works. In general, I tend to stay away from traditional multi-point autofocus of a camera because it focuses on the wrong subject way too often. However, I believe the Panny's new multi-point focus system allows you to use the touch screen to choose multiple focusing points, so I am really curious to know if this feature is intelligent enough to alter aperture selection based on your selected points of focus.

    Perhaps other Panny G3 or GF3 users (don't know if they have this feature on the G2/GF2 or not) can shed some light on this?
  4. Graywolf

    Graywolf New to Mu-43

    Aug 5, 2010
    Howi, Really, bad day, bad night, It was just a simple question.
    • Like Like x 1
  5. fredlong

    fredlong Just this guy...

    Apr 18, 2011
    Massachusetts USA
    Real Name:
    Multi-point focusing chooses from multiple points using various algorithms. Depending on the camera it may use one or several points, but they will all be in the same plane. It will not try to increase depth of field.

    If you choose a point on a Panasonic camera it will use a small cluster of three to five points to focus.
    • Like Like x 2
  6. Alanroseman

    Alanroseman Super Moderator Emeritus

    Dec 21, 2010
    New England

    Switch to decaf..... :biggrin:

    • Like Like x 1
  7. Luckypenguin

    Luckypenguin .

    Oct 9, 2010
    Brisbane, Australia
    Real Name:
    Canon has a mode on their DSLRs called A-DEP (Auto depth of field, I think) which does exactly what you describe, but I don't think the Panny multi-point focus is anything more than letting the camera choose it's own focus point.
    • Like Like x 2
  8. nickthetasmaniac

    nickthetasmaniac Mu-43 All-Pro

    Jan 11, 2011
    My understanding is that all of the focus points the camera selects when using multi-point will be in the same focal plane.

    That is, it just pics up all the things that will be in focus. It doesn't make additional things go in focus.
  9. Howi

    Howi Mu-43 Veteran

    Feb 23, 2011
    Real Name:
    Yup! sorry folks, bad day......
    ignore my rantings, hopefully tomorrow will be better :smile:
    • Like Like x 2
  10. GaryAyala

    GaryAyala Mu-43 Legend

    Jan 2, 2011
    Forget decaf Howi, relax with a glass of Zinfandel.

    New Daddy, typically, when using multi-point focusing, the points in use will focus (defer) to the object closest to the camera and that focus point.

    I prefer to use center focus point only and switch my focus activation from the shutter release to the AF/AE Lock button on the back of the camera. That way I can focus, release the button (the focus locks) re-frame my image then release the shutter without losing my focus.

    The only adjustment which will alter DOF is aperture. The higher the F/Number the greater the DOF. With aperture adjustment (increasing the F/Number), DOF increases more behind the focus plane than in front of your focus plane. The best way to affect aperture is probably in "A", "S" or "M" mode. (Possibly "P" as well but to be honest I've never have use "P" mode so haven't a clue to how it works.) To increase the DOF, after a baseline exposure setting, one must either decrease shutter speed or raise ISO or both.

    • Like Like x 1
  11. WoodWorks

    WoodWorks Mu-43 Regular

    Oct 10, 2011
    Ashland, OR USA
    Real Name:
    I'm curious about this too. But specifically, what I want to know is this: When using face detect focus, and the faces that the camera detects are at considerably different distances from the camera, does the camera adjust the aperture to accommodate those different distances? It seems to me this might be a different scenario from multi-point focus, because the face detect focal points would be more specific. Any guesses? I wonder if this might be fairly easy to confirm by lining up a few willing victims, er, friends. :rolleyes:
  12. New Daddy

    New Daddy Mu-43 Regular

    Jan 24, 2011
    If the multi-point focusing only picks up the objects in the same focal plane, I don't think it would be much different with the face detection: it probably picks up the faces in the same focal plane. I'm going to run some trials to make sure though.
  13. New Daddy

    New Daddy Mu-43 Regular

    Jan 24, 2011
    I prefer center focusing too, and as a matter of fact exclusively use center focusing.

    My question arose because, with center focusing, it was really difficult to see whose faces were in focus, especially in the absence of any help from a DOF scale. Without knowing whether and by how much I should decrease the aperture (i.e., increase the DOF), it practically became a hit-or-miss. The problem was exacerbated as it was a low-light indoor shooting, which required the largest aperture feasible. I had to chimp to make sure that all were in the focal plane, but even with the 4x magnification on the LCD, it was hard to decide one way or another.
  14. starlabs

    starlabs Mu-43 Top Veteran

    Sep 30, 2010
    Los Angeles
    I'd like to know what the behavior is with multiple face detection as well.

    If the algorithm/camera assumes they're all on the same focal plane, well that's kinda not so useful (great for single person portraits though)
  15. phigmov

    phigmov Mu-43 Hall of Famer

    Apr 4, 2010
    Completely agree - I'm guessing all the camera will do is shut-down the aperture until all the faces it detects are in focus.

    Downside is the more thats in focus the less the people will stand-out from the background but thats a fairly immutable optical thing anyway.
  16. shnitz

    shnitz Mu-43 Top Veteran

    Aug 25, 2011
    Austin, TX
    Multi-point basically is the same as fully auto, but you select the area, in case of complex scenes. It will choose the best focus point, which is most likely the closest. The reason for this mode:

    1. Let's put the camera on full auto, and take a picture of our friend, standing in front of the camera, in a wide open field. The camera will focus on our friend, and take the photo. Specifically, it will choose the focus point on her that is closest to us, and it will adjust the point of focus to that individual distance.

    2. Now, for a more complex scene. Have her stand just to the left, and two feet back, from a tree trunk. Keep the camera on full auto again, frame the trunk and our friend in the photo, and the camera will choose to focus on the tree trunk, since it is the closest subject in the frame. This is obviously bad, because we're trying to take a portrait, not a picture of trees with blurry people behind them.

    3. Let's fix this using multi-point focusing. Now, we get to select the area for the camera to concentrate on. We select the area of the photo with our friend, which gives the camera insight into the important subject in the frame. The camera will choose the point of focus within this area, and set the focus to that distance.

    This is a focus mode, and it has absolutely nothing to do with the depth of field. DOF is purely a function of aperture, lens focal length, and subject-to camera distance. Point of focus can only be set to 1 distance; it's a function of the physics behind photography.

    Face detect works the same way: it will detect the most prominent face, and use that one as the point of focus. It will not perform some kind of magic. If you want to ensure that a group of people are all in focus, then turn on some lights to get some more illumination, set your aperture smaller to ensure a greater depth of field, and step back as far as you can, so that the various subjects are all about the same distance from you. If you are two feet away from one subject and 6 feet away from another, then the second person is 3x further away and will most assuredly be out of focus. But if you walk back 10 feet from the first subject, he is now 10 feet away and the second subject is only 40% further, at 14 feet. This will help you to capture your shot. Also, almost always focus on the closest person, because the DOF extends further behind your point of focus than in front of it.

    There's a ton of information on the internet. Use google, plus some good search words, to find all the explanation (and mathematics) behind DOF that you'd ever want to know.
  17. Howi

    Howi Mu-43 Veteran

    Feb 23, 2011
    Real Name:
    As cameras become more sofisticated, the less we know about HOW they actually work.
    This is where the manufacturer could really help by explaining what the various algorithms are.
    Multi point focus, face detect etc all rely on following a set of rules determined by the software designer.
    try a simple test, put focus on multi point auto, choose a scene with no particular subject in the middle of the screen (landscape type scene) half press the shutter button and see how many focus points appear, release and do the same again - do you get the same focus points on screen?
    do the same with face detect on, choose a scene with a number of faces in it, which ones are detected? (refocus and see if any different?)
    Faces can now be remembered and checked against the faces detected with names being applied on screen so there is some serous processing going on here.
    Now, getting down to single focus point which (i think) most of us use by default.
    Take the standard sized focus point and try and focus on something, how many times is the object you require, in focus?
    Well, this will be determined by lots of factors, but in general is very hit and miss - cue now for getting the smallest focus point we can, we now have more chance of getting our primary subject in focus. Why would this be?

    I am not sure about multi focus points or face detect, but for single focus point on a Panasonic GF1, it would start with the lens at infinity and bring the focus forward until contrast is detected (I presume between any adjacent pixels within the area defined by the focus square)
    This method will always tend to focus behind the subject, hence selecting a smaller focus area reduces the problems quite significantly.

    New Daddy has certainly raised some interesting points, unfortunately these can only really be answered by the specific camera manufacturer. I doubt if any would be willing to give away their trade secrets though, which doesn't hep us much.
    He also raises a very interesting point regarding face detect being linked into exposure (aperture specifically) to ensure all faces are within the depth of field, but this leads on to what is the minimum size of face that can be detected(and on what defines a face).

    I am sure this facility will be available soon, but unless there is plenty of light available could be self defeating - increasing the aperture to say F8 or F11 to get all in focus could lead to very slow shutter speeds increasing the chances of motion blure/camera shake which would negate the original objective.

    This will always be the problem of letting software (no matter how good) make your photographic decisions for you, it is always much better, armed with the approriate knowledge, to make those decisions yourself.
    In good light, it would be very easy to do with success more often than not, in lower light it gets more challenging to the point of not being possible at all (get down to an aperture of F1.7 and see how many faces you can get in focus.....
  18. New Daddy

    New Daddy Mu-43 Regular

    Jan 24, 2011
    I think your statement is just going in circles. Nobody is denying that "DOF = fx(aperture, focal length, subject distance)"

    What the rest of us are discussing here is whether our Pannys can set the aperture automatically based on the focus points that the user has chosen. It perfectly complies with the law of physics and mathematics. From the above equation, if you are given the DOF, focal length and the subject distance, you can certainly calculate the aperture through reverse calculation. Or even better, you can make your CAMERA do the calculation for you. That's what is being discussed here.

    Obviously Canon does have such mode, although they don't seem to be well-received by the users. Check out the following links:

    Canon's Depth-of-Field Exposure Mode
    Peachpit: Canon Rebel T1i/500D: The Creative Zone > A-DEP: Auto Depth of Field Mode
  19. shnitz

    shnitz Mu-43 Top Veteran

    Aug 25, 2011
    Austin, TX
    There are no trade secrets or secret mumbo-jumbos occurring. We still know EXACTLY how cameras work; despite technology advancements, they're still bound by those pesky laws of physics. Read my post that's right above yours, but more slowly. The camera will focus on the closest subject, no matter which mode you're in. This has always been true for autofocus. The various modes just adjust which closest subject to focus on. There's no magic. It's all very simple equations. Go pick up a technical book about photography to understand focus points. If you choose face detect, it will focus on the closest/most prominent face, understanding that you chose people as the subject, so it will ignore subjects such as walls, tables, chairs, and trees. Focus mode WILL NOT AFFECT EXPOSURE SETTINGS. What defines a face is not voodoo. It is simply a set of algorithms:
    Face detection - Wikipedia, the free encyclopedia

    Now, if you choose multi point, then it focuses on the closest subject within a certain portion of the frame. If multiple focus points are shown, then the camera has chosen one, and is just showing you other subjects that are the same distance from the camera. Even a cursory glance in your manual will show you this.

    Here is from my G2 manual, page 79:
    "When the camera detects a person’s face, the following color AF area is displayed.
    Yellow: When the shutter button is pressed halfway, the frame turns green when the camera is focused.
    White: Displayed when more than one face is detected. Other faces that are the same distance away as faces within the yellow AF areas are also focused."

    I bolded and underlined the relevant statement, to ensure reading comprehension. The camera, whether it's a disposable film camera from Wal-mart, a GH2, or a $60,000 medium format camera with a digital back, focuses to ONE distance.

    You don't have to stop down quite as dramatically as you claim to get a solid depth of field. Go look at an online DOF calculator to get an idea of what your camera is doing, since m4/3 lenses don't have DOF scales on them.
    Online Depth of Field Calculator

    As an example, I put in a 20mm lens, focused 10 feet away at f/4. The depth of field is from 7 feet in front of you to 18 feet in front of you, giving you 11 feet of DOF to get the table in to. Open the aperture up to f/2.8, and the depth of field is still from 7.5 to 14 feet away, giving you 7 feet. No magic necessary, no trade secrets required. Use proper technique, and a tiny bit of thought.

    Put your kit lens on, zoom out to 14mm, and shoot it at f/4. Focused 10 feet away, your depth of field is now from 5.5 feet away to over 50 feet away. Not so hard, was it?
  20. shnitz

    shnitz Mu-43 Top Veteran

    Aug 25, 2011
    Austin, TX
    Let me boil it down, last time, just in case someone doesn't want to read all of that: Face detect, multi point AF, or any other focus mode WILL NOT adjust anything besides the distance that the lens is focused. That is absolutely the only thing it does. It doesn't do anything to the aperture.