Announcement

Collapse
No announcement yet.

Raspberry Pi Announces The $50 High Quality Camera

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #51
    Originally posted by oiaohm View Post
    Really you cannot compare a smartphone main camera to one designed for a changeable lens. The silicon area of the sensor for one made for a lens is on a totally different scale and totally different production.
    ? They're often 1/3" or 1/2.5" no? Doesn't really matter, you can get a suitable lens and mount, these image sensors aren't exclusively for mounted lenses. Smartphone products can often be found with high quality optics on the premium end that you won't be finding in the budget camera modules available to public. Smartphone cameras often have other features/integrations beyond the raw sensor specs, which again isn't something we will come across easily due to all the NDA / IP behind that.

    Even if we can get the sensor, to make our own module requires a fair amount of money for licenses/memberships to implement stuff like MIPI-CSI properly and interface with the camera, so as that's not really realistic, we have to hope for the best through companies like RPi Foundation that fund such by creating products with it.

    There are changeable lenses that can be mounted onto the back of smartphone cameras btw. Due to the other advantages these sensors have, the drawback of layering another lens on top isn't as bad as it sounds, you can still get quality results compared to what is usually on offer for hobbyists.

    Originally posted by oiaohm View Post
    You have a mistake here the new Pi camera is not using custom designed lens neither was the prior Camera Modules.
    No, see the blog post.. https://www.raspberrypi.org/blog/new...ale-now-at-50/

    >
    As with every product we build, the High Quality Camera has taught us interesting new things, in this case about producing precision-machined aluminium components at scale (and to think we thought injection moulding was hard!). Getting this right has been something of a labour of love for me over the past three years, designing the hardware and getting it to production. Naush Patuck tuned the VideoCore IV ISP for this sensor; David Plowman helped with lens evaluation; Phil King produced the book; Austin Su provided manufacturing support.

    > We’d like to acknowledge Phil Holden at Sony in San Jose, the manufacturing team at Sony UK Tec in Pencoed for their camera test and assembly expertise, and Shenzhen O-HN Optoelectronic for solving our precision engineering challenges.

    The "lens", as in the in optics was evaluated among many to find ones that are best suitable for their product. But the barrel according to that is custom designed to accomodate components like the chosen optics. Not unlike an SBC selecting an ARM processor from another vendor, but the board itself is the product and designed by the team, sourcing components where necessary, but still a custom product at the end.

    They did not just take an existing lens product and decide to sell it at a higher margin under the RPi brand.

    Originally posted by oiaohm View Post
    Yes the raspberry pi Camera Module V2 is in fact CS lens compatible as well but requires CS lens direct mounted and you have to remove the provided lens.
    You can find the IMX219 with CS lens/mount. Arducam the provider you linked earlier provides such: https://www.arducam.com/product/ardu...a-v2-cs-b0102/

    So I'm not sure what you're on about? You can also take a board with just the sensor and add your own mount holder, in this case CS-mount, just screw in:
    https://www.aliexpress.com/item/32665147668.html

    Nothing new there, but the IMX477 being available and supported is nice, the added CS-mount pre-installed is also a nice bonus.

    Originally posted by oiaohm View Post
    Little point something with a 16-30 MP sensor normally cannot do 30 frames per second video.
    ? What are you trying to go on about here. Those Pentax cameras are superior to any of the pi stuff. Not just because of MP value, they have pixel pitch almost 3x the size, you'll get a far better image.

    Video wise, they're a bit old and can only manage [email protected] or [email protected], this is the same restrictions on the IMX477 when used with any RPi board.

    Originally posted by oiaohm View Post
    12-13MP is 4K and your 35+MP sensor is your 8K and there appears to be a sensor black hole in the middle for any decent speed capture.
    4K is roughly 8MP, 8k roughly 33MP. No idea what you're on about with a "sensor black hole" or why 8k capture is coming up, even with a proper board to support the IMX477, it's only capable of [email protected] at best, and that's not even 5K resolution, so where is 8k coming from?

    Originally posted by oiaohm View Post
    Yes the one the new Pi high quality pi camera is using is the 30 frames lower end at 4K capture.
    Not sure what you're trying to say here. It can do [email protected] 10-bit(SME-HDR) capture on hardware that can support it. RPi boards are incapable of that, max they support is [email protected], you could take 4k stills and perhaps get 5 or so frames a second? (random number, not sure what you'd manage on the pi)

    Originally posted by oiaohm View Post
    polarathene I know about this speed of capture problem because a relation I have likes taking flying aircraft a few frames per second don't cut it you need a camera that is able to do 30 fps at least to be able to rapid fire and get aircraft in shot at times. So its not just a resolution problem of the sensor sometimes a camera with a low resolution sensor is a better choice than a higher resolution sensor.
    You're not making sense here either. You can capture different resolution with the image sensors, and each has different performance capabilities, if anything a larger sensor is more useful due to the larger pixel pitch to receive more light.

    I'm not sure what you're comparing to at this point, nor what you're referring to by 30FPS vs rapid fire, that depends on the resolution capture and sensor performing it.

    The perks of the small sensors are, weight an size which allows for use in smaller form factors. The lower power requirements, etc.

    If you want to capture at a higher frame rate, then select capable hardware. That IMX477 sensor can manage [email protected], just not on the RPi(30FPS max for 1080p). But for the use case you mentioned, you'd probably also want to prefer a sensor with global shutter instead of rolling shutter if quality at high speeds is important(unclear what you're taking a photo of, if it's the landscape below, probably non-issue, if it's another fast moving object, go for global shutter).

    Originally posted by oiaohm View Post
    I do think this raspberry pi high quality camera for what it is not badly placed. Something that is designed to connect to what you will be using in education or lab setting providing a reasonable resolution with usable capture speed.
    I've not said it's bad. It's just the lenses don't sound like they'll be what people expect to get out of the camera, they'll need a third-party lens instead. The limitations of the RPi board also make it a little less appealing, but I'm glad to see RPi Foundation put this product out, I hope it sees broader adoption/support in other products and linux so it can be used with more capable hardware.

    Comment


    • #52
      Originally posted by uid313 View Post

      Yeah, but they're still pretty shitty. They don't do 4K, HDR, 60 fps with 240 fps slow motion capture. The webcams kind of seem suck in the past and is way behind what is available in phones.
      Why would you want 240fps slow mo on a webcam? (assuming you're using it for video calls)

      Those that get fixed into a laptop display, just don't have much practical value beyond capturing your face in a stationary position. 4k displays still isn't all that standard beyond perhaps Apple users(maybe? I dunno) and gamers, rarely adds much value for the purpose of video calls either, other than taking up more processing power and network bandwidth, HDR is nice, dunno if 60FPS really matters much(many movies we watch are still 30FPS afaik).

      Nice stuff, but not really that important, I'd rather the budget goes elsewhere in laptop products. For dedicated USB webcams, sure, those could probably do with using a better sensor. If you want a 4k webcam (dunno about 240 fps slow mo..), go look at the logitech brio. It has HDR 4k at 30FPS, otherwise 1080p a 60FPS.

      Comment


      • #53
        Originally posted by polarathene View Post
        So cheaper and better. Both sensors are capable of auto focus btw, so not sure what you mean by fixed focus?
        Auto focus and fixed focus don't mean the same things.

        When I am saying fixed focus I am talking about what is in front of the sensor the lens setup. Yes you can add software chip after the sensor to process what the sensor provides to give some amount of autofocus but its never as good as if the in coming light was perfectly focused on the sensor.

        Originally posted by polarathene View Post
        The arducam you linked only mentions it as programmable and lacking support for things like 3A algorithms.
        3A algorithms are used two ways.
        1) in software autofocus that the arducam lacks
        2) for auto drive focus on a proper lens camera this is like you m12(complex versions) or CS or C or SLR lens.

        Originally posted by polarathene View Post
        They also offer m12 lens mount version for $3 more:
        Problem is there are a issues with the m12 mounting. There are 15-30 dollar m12 lens.
        http://www.m12lenses.com/M12-Fixed-I...enses-s/44.htm
        These do have focus and zoom rings like a CS or C mount lens.

        But there is a problem with the "adjustable back focus". In your cheaper m12 lens focus and zoom basically linked into 1 so this is not a problem. When you use the more expensive m12 lens this problem comes out. So the back focus on the more expensive m12 lens is done by the m12 thread. Can you see problem its simple really. Where is the "adjustable back focus" locking bolt or locking mechanism its missing off the m12 mounting so as you are adjusting the focus/zoom on the more expensive m12 lens you can also be mucking up the back focus.

        Remember not all lens are perfect so when you hit a m12 that you cannot screw all the way in because needs some back focus adjustment and you attempt to adjust the front you are now in a losing fight. If the budget in security is enough to ditch a m12 with a complex lens you do. Please note the vibration from power drive m12 lens also cause that back focus to drift out of focus to the point people give up having those cameras perfectly in focus.

        Originally posted by polarathene View Post
        With the RPi HQ cam, they offer a 6mm CS-mount lens or 16mm C-mount lens, here's a 5-50mm varifocal lens with camera module and USB(albeit it's not interchangeable for different lens and it's an 8MP module, IMX179..):
        arifocal-Megapixel-3264X2448-Definition-Industrial is in fact a interchangeable CS lens problem is in the mount. Why it cheap is they screwed up.by missing standard define focus adjust ring between the sensor and the lens is missing from the mount this is called "adjustable back focus". So it only works correctly with selected CS and C lens even of the same model lens. Yes you buy 2 the same lens and one would work on the camera you found and one would not due to minor manufacturing differences that you over come on a good mount with a back focus ring..

        The RPI HQ cam has all important back focus ring behind lens in the mount so can take fairly much every CS and C lens on the market with the correct image format and make it work right. So there is a key feature you want on a CS or C mount if it not there you are in trouble.

        Originally posted by polarathene View Post
        For $76, that isn't all that bad of an example price wise(few comments indicate yellow tinge, so probably not a great buy).
        People have bought it without knowing why it insanely cheap. That missing adjustable back focus means that you can be constantly running in software auto focus with all the artefacts that can cause. Its not suitable to use on a lot of Microscope because of the missing back focus ring. Something with a CS mount preassembled under 150 USD is not normally on special there is normally a screw up on it somewhere and they are trying to sell the inventory to what every suckers will take it.

        Originally posted by polarathene View Post
        Why do you need a RPi 4 to drive this camera?
        Its for the USB 3.0 you can send data to the RPi 4 USB 3.0 without stall out risk of the old USB 2.0. On encode I was forgoing how weak the RPi4 was. I have been using a Rock Pi recently vp8 encoder on it can be pushed out to 4K I was forgetting RPi don't have this option.

        Originally posted by polarathene View Post
        Note, I've looked into the RPi lenses with the HQ cam and they affect the Field of View and output resolution, you don't get 12MP with either of them. The 6mm lens offers FoV of 63 @ 3MP, while the 16mm varies from ~20-45 degrees @ 10MP. The 6mm is intended for macro photography and the 16mm for long-range/zoom usage, so neither is particularly useful for webcam usage?
        This does not read right at all CS or C lens will cover that complete sensor so 12MP is possible. Problem on these CS and C lens you will see what megapixel the lens was designed and built for not the mega pixel you can take though them. Image format is the key value. Image format 1/2" that the circle of the output of the len this is a little bit of a problem thinking the sensor is a "1/2.3-type" its inside the CS/C lens specification of tolerance on sensor placement so in theory you should with a 1/2.3 sensor with 1/2 CS lens be able to use all the sensor.

        Originally posted by polarathene View Post
        That gets clarified here(you could have a 12MP image, but quality degrades beyond the given range):
        This is not exactly true on the degrades once past the point of rated operation. The lens glass was not tested that it was perfect for a MP image past it rating is all it means not that is flawed past that point just it might be. Most people have heard of the silicon lottery by overclockers you also have the lens lottery you could get a 3MP rated lens that turns out to be a perfect 12MP lens heck even way past that. Of course that is more likely with a 10MP lens that it a perfect 12MP lens than a 3MP lens be a perfect 12MP lens. 25 USD CS lens I was thinking for is a 11MP rated security camera lens with a "1/2.5 image format the the tolerance that is normally a 12MP lens unless you are really unlucky. Here is a horrible on HD 4K security cameras factory supplied CS lens that work perfectly fine at 4K with no issue only have a 2MP rating and sometimes only a 1MP rating. So basically MP rating on CS and C lens is more a guild line than absolute rule its very possible once you know a vendor to be buying a 1MP rated lens for a 24MP sensor and know it will work perfectly fine. Lets just say security camera lens rating are no where near as nice as SLR lens ratings and the vendor making them don't really upgrade their testing gear to put higher rating on stuff yes they upgrade their lens production gear and produce better and better lens in their gear yet the MP rating on lens never shows this. So is nowhere near right its always under the question is always how much the most insane is a factor of 200 under rated so a 1MP lens in fact being constructed to 200MP quality does happen you just would not see this level of error in rating with a SLR lens.

        Basically I can understand the problem here reading security class C/CS lens ratings with a SLR lens mind set. The SLR mind set makes you think the rating on the lens is the upper the lens supports when that is just not the case with the C/CS lens. Working with security cameras you end up more often and not ignoring the MP rating instead being more worried about the image format as in will it use the full sensor yes/no.

        The 25 dollar lens I was thinking of is one I normally source that is fairly close to the lens that was on the webcam that has the bad CS mount.

        Originally posted by polarathene View Post
        Why are we bringing a motorized drive into this comparison and cost? It's a nice feature, but generally for a webcam, manual focus adjustment of the lens isn't a big deal due to typical usage,
        On good CS/C mounts its possible todo motorised driver and have no problems yes the raspberry pi HD camera mount is a good one. If you are going to have a proper CS/C mount with locking bolt you might as well take advantage of it stability and drive the heck out of it.

        Comment


        • #54
          Originally posted by oiaohm View Post
          Auto focus and fixed focus don't mean the same things.

          When I am saying fixed focus I am talking about what is in front of the sensor the lens setup. Yes you can add software chip after the sensor to process what the sensor provides to give some amount of autofocus but its never as good as if the in coming light was perfectly focused on the sensor.
          Not clear what you're saying here. ISP is the software chip? Those are ASICs specially designed to perform such functions afaik, they're doing the math you'd find anything else that does it in premium gear with autofocus I'm sure. That'd be firmware/hardware based.

          The $19 IMX214 module I linked, has a micro-motor for performing autofocus, ISP should be able to control that afaik. If you're using the camera in a fixed position with a fixed focal point, for sure ignore that and add a fixed focal length lens, autofocus isn't relevant then?

          I didn't say the meant the same thing btw, you called the arducam fixed focus, I pointed out that the sensor (IMX214) as a module like used in that product and the one I linked without the m12, has auto/adjustable focus, it's not constrained to a fixed focus. Perhaps the point you were trying to make was that it did not support inter-changeable lenses to change the focal length?

          Originally posted by oiaohm View Post
          3A algorithms are used two ways.
          1) in software autofocus that the arducam lacks
          2) for auto drive focus on a proper lens camera this is like you m12(complex versions) or CS or C or SLR lens.
          I was referring to the camera module without m12 lens, that has autofocus via 3A. I'm not 100% sure if that is only available via ISP, as you can see the Arducam product doesn't offer them, only way to control focus is via explicit control of the motor apparently? I assume because it does not have access/support to the RPi GPU ISP block that RPi products do or something?

          There are some other products for that label themselves as compatible with the Pi boards, but are just USB cameras with their own ISP integrated to handle processing, allowing them to deliver better performance than any camera you'd have over MIPI-CSI to the board directly(even when the GPU ISP is supporting it).

          To adjust the focus, my understanding is you need to move the optical lens, either changing distance from sensor, or physically moving the whole thing to adjust focus on the target. The version of the product I linked with the m12 lens, might provide manual adjustment, but it's not going to do autofocus like the version without m12. Sure, you can get motorized m12/C/CS/whatever lens and optionally control that via software to refine focus too.

          Originally posted by oiaohm View Post
          Problem is there are a issues with the m12 mounting. There are 15-30 dollar m12 lens.
          http://www.m12lenses.com/M12-Fixed-I...enses-s/44.htm
          These do have focus and zoom rings like a CS or C mount lens.
          Those are cool, thanks for pointing them out.

          Originally posted by oiaohm View Post
          The RPI HQ cam has all important back focus ring behind lens in the mount so can take fairly much every CS and C lens on the market with the correct image format and make it work right. So there is a key feature you want on a CS or C mount if it not there you are in trouble.
          This does seem to be a nice feature, thanks for mentioning it. Looking into it, some products provide automatic back focus features for their cameras, not something these ones I've been linking to would have of course.

          Still, it doesn't mean you are necessarily in trouble, only if there is some incompatibility preventing being able to achieve focus.

          Take the m12 lenses site you linked, it sells CS-mount lens holders, these don't have that screw, but they do have some sort of hole in the side(whatever that is for): http://www.m12lenses.com/CNC-Machine...t-lh026csm.htm

          I did find some additional specification information from Seeedstudio on the HQ cam: https://www.seeedstudio.com/Raspberr...am-p-4463.html

          > Back focus - Adjustable (12.5 mm–22.4 mm)

          I find that interesting, CS-Mount is 12.5mm flange distance, so this adjustable back focus ring is only helping to extend the distance? (C-Mount being 17.5mm via the adaptor adding 5mm spacing ring). That implies additional spacing ring of an increased length would work fine here too, even more so with C-Mount due to adapter.

          Originally posted by oiaohm View Post
          People have bought it without knowing why it insanely cheap. That missing adjustable back focus means that you can be constantly running in software auto focus with all the artefacts that can cause. Its not suitable to use on a lot of Microscope because of the missing back focus ring. Something with a CS mount preassembled under 150 USD is not normally on special there is normally a screw up on it somewhere and they are trying to sell the inventory to what every suckers will take it.
          I don't follow sorry. That $76 camera was an all-in-one package. I don't believe the lens was meant to be swapped out or anything. The issues people had with yellow tinge is presumably from issues with the optics quality or it's image processing, supposedly that can be resolved in post via software to some extent.

          Originally posted by oiaohm View Post
          Its for the USB 3.0 you can send data to the RPi 4 USB 3.0 without stall out risk of the old USB 2.0. On encode I was forgoing how weak the RPi4 was. I have been using a Rock Pi recently vp8 encoder on it can be pushed out to 4K I was forgetting RPi don't have this option.
          ...? You do know the RPi HQ cam is over MIPI-CSI2, not USB right? If you're just wanting to use the camera, the cheaper Zero boards are presumably adequate at handling the [email protected] h264 encoding, or whatever stills you want to take.

          Originally posted by oiaohm View Post
          This does not read right at all CS or C lens will cover that complete sensor so 12MP is possible. Image format is the key value. Image format 1/2" that the circle of the output of the len this is a little bit of a problem thinking the sensor is a "1/2.3-type" its inside the CS/C lens specification of tolerance on sensor placement so in theory you should with a 1/2.3 sensor with 1/2 CS lens be able to use all the sensor.
          Just going by what I've read, not an expert on the matter so I can't say. The MP value from what I've read isn't particularly useful, but lp/mm is when rated with the pixel pitch. The basler lens I later linked to (machinevision link) is rated at 230 lp/mm for 2.2 micron pixel size. I see your response didn't quite get that far, so perhaps you hadn't read it by this point. I explain the math there of how the 5MP rating is probably 8MP for the HQ cam.

          As for the the HQ cam lenses, I linked you to the blog post comment by the guy involved on the product (James). So he probably knows what he's talking about, doesn't help that the lenses are lacking other useful specs. You still get the 12MP if you want, but the quality degrades(as in blurs from diffraction). The 10MP seems understandable and not so bad, but that reduction to 3MP for the 6mm isn't particularly great sounding..

          Originally posted by oiaohm View Post
          Problem on these CS and C lens you will see what megapixel the lens was designed and built for not the mega pixel you can take though them.
          Well... as these are designed and built over 3 years for this specific product... and seeing as how RPi are selling them for their product.... one would assume that the lenses specifically mentioning the MP rating, and the linked comment I provided are proof enough that the effective resolution(sharp/crisp) when used with those lenses is what they're valid for. Going higher than 3MP would provide no beneficial detail then.

          As I don't fully understand this impact myself, it's unclear if they refer to the range such as 3MP as only 3MP/12MP of the sensor coverage, or as in the full sensor is covered, but a 3MP image is 1 pixel for every 4 pixels, like scaling down an image. That there is no notable quality/detail to be gained when sampling from a higher resolution. This seems more likely to be the case when matched with the lenses being rated at lines per mm.

          That's how I get the 8MP value for the basler lens, assuming the amount of "lines" per mm from the lens is fixed, there's only 8MP of detail available at that resolution for IMX477, to get higher detail than that would require more lines.

          Originally posted by oiaohm View Post
          This is not exactly true on the degrades once past the point of rated operation. The lens glass was not tested that it was perfect for a MP image past it rating is all it means not that is flawed past that point just it might be.
          They stated they did lens evaluation (on the optics) while designing the actual lens(unclear exactly how much they designed here, but it seemed to be in partnership with other companies with the expertise, still a custom design/product). Pretty sure if they're going to make a product for 12MP sensor, that they're not going to think it's a great idea to sell a 3MP lens unless there is some technical or cost issue around that. The ratings will be targeted at those lens and the sensor they're being sold with by RPi Foundation.

          There is little to debate otherwise, especially when they've confirmed the drawback.


          Originally posted by oiaohm View Post
          25 USD CS lens I was thinking for is a 11MP rated security camera lens with a "1/2.5 image format the the tolerance that is normally a 12MP lens unless you are really unlucky. Here is a horrible on HD 4K security cameras factory supplied CS lens that work perfectly fine at 4K with no issue only have a 2MP rating and sometimes only a 1MP rating.
          Is the "Here" meant to have some sort of link? All I'm getting without one is, you have a 4k security camera and lens, the 2MP rating is unclear though(especially since you later state "and sometimes only 1MP rating"), which suggests the product itself was stating that, rather than the MP rating being attributed to the lens itself. In which case, it may very well be a 1-2MP sensor and upscaling to 4k, I don't know because I don't have other information to go by, other than it's supposedly delivering 4k but it's horrible, whlie the lens itself is supposedly fine.

          To further add to that, the lens could very well be rated at 2MP but you need to take the actual lp/mm value into account afaik and compare that to the sensors lp/mm, so on a different sensor, sure perhaps if lens is actually good and sensor isn't, then you get better quality on better sensor. Same with bad lens on good sensor producing bad result.

          Originally posted by oiaohm View Post
          the vendor making them don't really upgrade their testing gear to put higher rating on stuff yes they upgrade their lens production gear and produce better and better lens in their gear yet the MP rating on lens never shows this. So is nowhere near right its always under the question is always how much the most insane is a factor of 200 under rated so a 1MP lens in fact being constructed to 200MP quality does happen you just would not see this level of error in rating with a SLR lens.
          That's probably because the MP rating really doesn't matter, megapixels is a useless metric for lenses from what I've read. Proper testing is using MTF charts, and providing a lp/mm value @ 20% contrast.

          If you're not familiar with what I'm talking about, and don't really want to read up about it, it's a metric based on how many lines the sensor can read of alternating black & white stripes/lines until the gap between the black lines gets too small that the sensor cannot make an accurate distinction(think of it as binary 0/1, instead of 1010101, ends up being 1010111). This is still partly tied to the camera sensor being tested with the lens however, so yes, there's probably some leeway, but from what I understand, it's fairly reliable metric when the vendor actually publishes it with the tested camera pixel pitch (like the basler lens I linked, not all of their lens include that information).

          This is also known as the lenses resolving power.
          Last edited by polarathene; 03 May 2020, 09:13 AM.

          Comment


          • #55
            Originally posted by polarathene View Post

            Nah, not really 😅

            I'm just a hobbyist on a budget so I don't pay the nice cameras too much attention, I mentioned the Sony A7R IV because my photogrammetry buddies were quite excited about it when it was making the rounds. I'd like to do some DIY embedded stuff with the not so great image sensors/lenses, one day if I can afford to get a sweet DSLR, I will
            That describes me too.

            Quick comparison after looking at the links for K-70 and K-S2, used this:
            https://www.imaging-resource.com/cam...s/pentax/k-70/

            The K-70 seems nice, but not really worth over 1.5x the price? I guess it depends how long you see yourself with the purchase until replacing it. I'd buy neither personally since I'm broke AF 😂 but if I could afford it, the K-S2 makes more sense, the 96MP composite image sounds cool, but I dunno how gimmicky that is, if you think it's something you'd care about and can try it out for yourself or something, maybe that and the additional ISO range is worth throwing the extra cash for?
            The price isn't that fair on that site. It's comparing the K-S2's base price with the K-70+ a lens that costs upwards of $450 when bought alone. The K-S2 with the same lens is usually priced within $100 of the K-70...that's the used on Amazon K-S2 price mind you...except for one single listing on Amazon where it costs almost the same as the base model and is new. How odd, huh?

            Base model to base model, usually around a $50-$100 price difference between the two. K-S2 is a 5 year old, discontinued model, so it's getting harder and harder to find. I am actually surprised that Amazon listing is a cheap as it is...very, very tempting.

            Actually, I wrote the above yesterday and got distracted with a bid I had to go on, checked prices today and the K-70 with that lens is $6 more than the used K-S2 with that lens (went on sale at B&H) so I'm basically debating between the old model for $556 or the upgraded model for $706.

            I'm intentionally paying more because it comes with a quality lens. A decent lens seems to be the one thing, regardless of brand or camera, that is universally recommended and that the first thing a person should do is upgrade from the low-end kit lenses most entry level cameras come with. Figured I'd pay a bit more now over trying to justify buying a $450+ lens later.

            I'm also broke AF and normally wouldn't consider spending that kind of money on a camera if it weren't for something like the stimulus check and that I can hopefully use it as a tool to make money. Due to what I do, residential contractor, my work has actually slightly increased since COVID so the stimulus check is just free money for something cool. All these people stuck at home looking at their shit that needs fixed is helping me out

            Comment


            • #56
              Originally posted by skeevy420 View Post
              Base model to base model, usually around a $50-$100 price difference between the two. K-S2 is a 5 year old, discontinued model, so it's getting harder and harder to find. I am actually surprised that Amazon listing is a cheap as it is...very, very tempting.

              Actually, I wrote the above yesterday and got distracted with a bid I had to go on, checked prices today and the K-70 with that lens is $6 more than the used K-S2 with that lens (went on sale at B&H) so I'm basically debating between the old model for $556 or the upgraded model for $706.
              Oh... good catch there with the price comparison. Well then... if there's not that much of a savings ratio from going with the old gear, might as well just get the newer one?

              Looks like you can choose between the body+lens for K-70 at ~$660 35mm or 50mm lens, ~$800 18-135mm.

              No input beyond that from me as I have no freakin' clue what one is worth choosing. Those fixed focal lengths from what I understand would still allow you to adjust focus in some way? The varifocal 18-135mm sounds nice, which afaik equates to wider to narrow FOV, or as an optical zoom..? If you're going to do real estate photography, might as well get those "spacious" cheat shots with a wider angle taken from the very corner of the room? :P

              Sounds like you're doing ok, I'd probably go for the more flexible lens for a little bit more then. Not that I make the best financial decisions

              Comment


              • #57
                Originally posted by polarathene View Post

                Oh... good catch there with the price comparison. Well then... if there's not that much of a savings ratio from going with the old gear, might as well just get the newer one?

                Looks like you can choose between the body+lens for K-70 at ~$660 35mm or 50mm lens, ~$800 18-135mm.

                No input beyond that from me as I have no freakin' clue what one is worth choosing. Those fixed focal lengths from what I understand would still allow you to adjust focus in some way? The varifocal 18-135mm sounds nice, which afaik equates to wider to narrow FOV, or as an optical zoom..? If you're going to do real estate photography, might as well get those "spacious" cheat shots with a wider angle taken from the very corner of the room? :P

                Sounds like you're doing ok, I'd probably go for the more flexible lens for a little bit more then. Not that I make the best financial decisions
                That's what I was thinking about the prices. Couldn't believe it when I saw the K-70 with the 18-135 for $706 earlier today. Hell of a deal. Only my dumbass read 18-50mm as 18-135mm so it is still an $800 setup . That's why I don't do important things first thing in the morning. I always mess up some mundane detail.

                What's crazy is the Pentax KP is down from the $1100-$1200 range into the $850 range with a 35mm or 50mm prime/fixed lens. I'm wondering if their prices are dropping because Pentax is supposed to be announcing a new DSLR soon.

                That's what the 18-135 does and it's supposed to produce some really good shots between the 25-120mm range. Since the 35mm and 50mm are in the $100-$150 range and the 18-135mm costs around $450, I figure that I should get the 18-135mm package since it's both a much more versatile lens and I can get some fixed prime lenses later on in the future since they're much more affordable.

                I've researched this subject a lot...only every time I've gone to pull the trigger in the past year and a half some sort of financial BS happened that wiped my savings so I kept putting it on the back burner. Since things are looking better for me now I figured I should get something nice and that I want while I have the chance. The world is on fire, but, oddly enough, I'm doing fantastic.

                Comment


                • #58
                  Originally posted by polarathene View Post
                  You can get similar if what you're after is 4k webcam, look at Logitech Brio:

                  https://www.logitech.com/en-nz/product/brio

                  It'll cost more, but that's because they're throwing in the enclosure / packaging / and all that other stuff to target a consumer. Might include some additional hardware for processing before sending over USB too.
                  Webcams don't have mounts for proper lenses, so you'll be stuck with tiny focal length that distorts the image and inferior optics.

                  Comment


                  • #59
                    Originally posted by DoMiNeLa10 View Post

                    Webcams don't have mounts for proper lenses, so you'll be stuck with tiny focal length that distorts the image and inferior optics.
                    Mod it mate: https://www.kurokesu.com/main/2017/0...-instructions/

                    You can pay the guy to do the mod for you if that's not your thing, but cost wise, kinda silly amount for a webcam.

                    Comment


                    • #60
                      Originally posted by polarathene View Post
                      The $19 IMX214 module I linked, has a micro-motor for performing autofocus, ISP should be able to control that afaik. If you're using the camera in a fixed position with a fixed focal point, for sure ignore that and add a fixed focal length lens, autofocus isn't relevant then?
                      To be correct fixed focus is wrong. Very limited focus and zoom combinations I think of them as fixed focus because you will be running up against the focus limits lot more.

                      Something you don't think of is your focus limit at times is defined by amount of light you lens area pulls in. Smaller you light capture less focus you can do.

                      Your bigger lens gives you more light capture more area to move lens for zoom and focus. Less needing to bend light angle as hard as well this make lens construction to get decent results simpler.

                      "ISP should be able to control auto focus" you hope so but its not always the case. There are some MIPI_CSI-2 camera with integrated auto focus that you cannot say stop at X focus. If the ISP is fully controlling the focus you are sure you can stop it where you want.


                      Originally posted by polarathene View Post
                      I was referring to the camera module without m12 lens, that has autofocus via 3A. I'm not 100% sure if that is only available via ISP, as you can see the Arducam product doesn't offer them, only way to control focus is via explicit control of the motor apparently?
                      You can do 3A focus in the camera module itself or have host/application side do 3A focus. Arducam put it on the RPi Cpu/GPU if you want 3A focus by software drive.

                      Yes even hardware driven lens by MIPI_CSI-2 you can be receiving messaging from the camera correct focus X direction please messages.


                      Originally posted by polarathene View Post
                      There are some other products for that label themselves as compatible with the Pi boards, but are just USB cameras with their own ISP integrated to handle processing, allowing them to deliver better performance than any camera you'd have over MIPI-CSI to the board directly(even when the GPU ISP is supporting it).
                      https://www.mipi.org/sites/default/f...tion_Brief.pdf

                      Sorry to say this is not true but there is a cost reason. MIPI_CSI-2(what is on all the RPi) is in fact a 6Gbps connection+. MIPI_CSI-2 does not forbid camera side processing. This is a link that can run rings around any USB 2.0 connection. Its just simpler to build a MIPI_CSI-2 to usb than what it is to build a MIPI_CSI2 to MIPI_CSI2 as you have to insert a middle processing item between camera and device and add driver to kernel to support this. MIPI_CSI-2 devices lots of them take firmware because they have processing in them.

                      To adjust the focus, my understanding is you need to move the optical lens, either changing distance from sensor, or physically moving the whole thing to adjust focus on the target. The version of the product I linked with the m12 lens, might provide manual adjustment, but it's not going to do autofocus like the version without m12. Sure, you can get motorized m12/C/CS/whatever lens and optionally control that via software to refine focus too.



                      Originally posted by polarathene View Post
                      Looking into it, some products provide automatic back focus features for their cameras,
                      Generally you cannot unless you put some form of calibration mark in the lens it self this is no longer generic lens and normally means losing some sensor space as well..

                      Problem is back focus screwed on the sensor also looks like front focus out. This is why you set and lock the back focus so you are only playing with 1 problem.

                      Originally posted by polarathene View Post
                      Still, it doesn't mean you are necessarily in trouble, only if there is some incompatibility preventing being able to achieve focus.
                      Unstable back focus makes it basically impossible to get front focus right. Remember changing back focus looks like changing front focus. So its quite important your back focus is stable(locked) and right.

                      Originally posted by polarathene View Post
                      Take the m12 lenses site you linked, it sells CS-mount lens holders, these don't have that screw, but they do have some sort of hole in the side(whatever that is for):
                      That is the shocking bad back focus lock design is that hole. You put a grub screw or equal in that hole that tightens down on the thread of the lens possible deforming it so the lens possible never comes back out and you can never adjust that back focus ever again if that happens you better hope you had it right.

                      Good looks like the raspberry pi HQ camera mount. So all CS mount lens expect you have some form of back focus be it that horrible grub screw or a proper locking bolt design that does not damage the lens thread. Yes if you force undo a CS mounted lens with a grub screw or equal lock you can basically de-thread the lens.


                      Originally posted by polarathene View Post
                      > Back focus - Adjustable (12.5 mm–22.4 mm)

                      I find that interesting, CS-Mount is 12.5mm flange distance, so this adjustable back focus ring is only helping to extend the distance? (C-Mount being 17.5mm via the adaptor adding 5mm spacing ring). That implies additional spacing ring of an increased length would work fine here too, even more so with C-Mount due to adapter.
                      Yes that is right its only to extend the mount. CS and C lens are fairly nice this way. If the focus length is wrong is normally long on C/CS lens. Why someone reads the C/CS specification sheet and they place the sensor too far back when making the lens or it been design for some camera with the sensor placed too far back.

                      You cannot go under 12.5 in a CS mount with safety other wise features sticking of back of some lens could possible hit sensor so cracking it.


                      Originally posted by polarathene View Post
                      I don't follow sorry. That $76 camera was an all-in-one package. I don't believe the lens was meant to be swapped out or anything. The issues people had with yellow tinge is presumably from issues with the optics quality or it's image processing, supposedly that can be resolved in post via software to some extent.
                      I have had someone buy that thing and bring it to me. That package started life being sold as split up parts. So Camera and Lens as two different parts before people started complaing. Later on they are packaged. Horrible part is they have been combined because the lens and camera body have a problem each. Optics on the lens results in incorrect focus length for a CS lens not far enough out to be problem if you have a back focus. Large percentage of the CS market ie security cameras don't want back focus because its a extra cost. Of course why this camera is reject as lab camera is no back focus to correct for bad setups and idea of digital auto sharpen(digital focus) always on that did not work perfectly to replace the lack of back focus. Don't have the camera here any more I found it from the part numbers that it was sold 12 months before with the same yellow tint problem.

                      Also just to make things a little worse the sensor in the camera is at the incorrect depth in the CS mount. In the case find the sensor in it is slightly too far forwards and the lens is focus long so you are in hell. So in theory a CS lens could have a bit sticking out the back and directly break the sensor in it(and you can guess why some of the original were returned). So something like this you are not going to advertise the changeable lens feature that much right.

                      So how do you attempt to hide that this complete camera is horrible out of focus on the sensor. Just like you can do digital zoom you can do digital focus maths correction and yes the errors in that shows up as yellow tinge few other wrong colours.

                      This is a case of priced too good to be true. Yes you get a camera and lens but they don't in fact work right. To fix the camera and the lens would be basically take both part and rebuild them right. So sell at material cost to get rid of them.

                      Originally posted by polarathene View Post
                      ...? You do know the RPi HQ cam is over MIPI-CSI2, not USB right? If you're just wanting to use the camera, the cheaper Zero boards are presumably adequate at handling the [email protected]S h264 encoding, or whatever stills you want to take.
                      I was not aiming to drive it that low but it would work.

                      Originally posted by polarathene View Post
                      They stated they did lens evaluation (on the optics) while designing the actual lens(unclear exactly how much they designed here, but it seemed to be in partnership with other companies with the expertise, still a custom design/product). Pretty sure if they're going to make a product for 12MP sensor, that they're not going to think it's a great idea to sell a 3MP lens unless there is some technical or cost issue around that. The ratings will be targeted at those lens and the sensor they're being sold with by RPi Foundation.
                      The issue here is technical you will find 4K sensor ~12MP security cameras with 1-3MP rated lens. MP rating on lens is on specification sheet and printed on does not mean much really of what the lens can do at all.

                      Originally posted by polarathene View Post
                      That's probably because the MP rating really doesn't matter, megapixels is a useless metric for lenses from what I've read. Proper testing is using MTF charts, and providing a lp/mm value @ 20% contrast.
                      There is no link between the MP value on the side of a CS/C/M12 mount lens or MP rating in specification sheet and the MTF chart of that lens other than that the MTF is better. MP value on these lens is so close to useless it not funny. So you can have a CS/C/M12 mount lens with 1MP printed on it and you look at the MTF past question its 200MP lens in it quality.across the complete sensor area sometimes this is a direct result of being design to support a broad range of cameras with different pitch sensors.

                      So you can basically say its a myth that you need a 12MP lens with a 12MP sensor to get good 12MP photo. You need a lens with a MTF rating suitable to the sensors pitch for it to work well to it full function and that is not printed on the lens you only find that out when you dig into the lens specification sheets and if you can get them in the first place.

                      Some will have the MTF rating of the individual lens that make up the lens but not the assembled rating as well.

                      Comment

                      Working...
                      X