[SOLVED] Image quality on Git2P after 90 FOV lens swap

I have no idea why they don't match, must be an explanation but unfortunately China is on holiday so it may be a while before you get an answer from Gitup. I can't see how using the wrong base will affect your image other than it may result in it going out of focus in very hot/cold weather, but unless you use it as a dashcam in parking mode then it probably wont make a noticeable difference.

Question is, is there any improvement having reseated it? But I guess we need to wait for daylight so you can focus it at a decent distance, no point trying to focus the 90 degree accurately indoors.

I fixed the links to the images as they were not showing up on the previous post.
The 90 FOV base is so off the mark that even if I wanted to use it, it'd be impossible to fit the screws as they don't line up with the holes on the mainboard.

I'll keep you guys posted.
 
Well yes, I've mentioned even in the thread's title that I have the Git2P -which obviously stands for Panasonic.
However, this exact model comes with a 90 degree 4.1mm lens which should be the one that the company sells on their website.
they sell a lens to suit the Sony version, it is not necessarily the same lens as what is used on the Panasonic version of the camera
 
I fixed the links to the images as they were not showing up on the previous post.
The 90 FOV base is so off the mark that even if I wanted to use it, it'd be impossible to fit the screws as they don't line up with the holes on the mainboard.

I'll keep you guys posted.
I had managed to see the full images anyway.

I suspect jokiin is correct about the Sony/Panasonic issue, but I think it will be the same lens, just a different base, and in any case, the focus issue across most of the image could not be caused by it being the wrong lens, if there was a difference it would only affect the very edge of the image.
 
When you install the 90deg lens in the 170deg base how are you securing the lens? I've swapped a number of lenses in various cameras and found the most secure way is to use a threaded locking collar, which not only keeps the lens at a particular focus point but also keeps it aligned with the holder. Before using locking collars I frequently had issues with lenses appearing to be out of focus on one side or another.
 
Tony TBH until you mentioned these locking collars and I googled them I had never even seen one in my life! :lol:
Def something to keep in mind.

So today I put back the 90 FOV lens and started focusing it more carefully.
I put the focusing pattern further away and then focused on it repeatedly, moving the camera so that the edges would be focused as well as the center FOV.

Then I took a sample pic outside.
This was the initial outcome.
Refocus sample #1

I was pleasantly surprised to find that the overall sharpness was much better than yesterday's pics.
Methinks it still left a bit to be desired though, so I went about fine-tuning the lens further.
This was the second result.
Refocus sample #2

What do you think? After a side by side at a 100% magnification I think the second pic is even sharper around the edges. (Check out the wooden balustrade and window frames of the house at the far right side of the image.) Nitpicking it could be ever so slightly softer on the far left side but it might be going too far.

So a final little tune up and setting the ISO to 50 with shutter speed at 1/125 resulted in this:
Refocus sample #3

What do you guys think? Is this an acceptable result? Should I call it a day?
 
What do you guys think? Is this an acceptable result? Should I call it a day?
Focus on the last link looks about perfect to me, but you could try a photo of something a bit closer before you pack up, you may find it wont focus at less than 10 meters now and you might want to adjust the compromise if you aren't just using it for landscapes.

There is rather more colour fringing in the trees than I would like, but that is not an installation issue and I think it is probably normal for that lens and wouldn't be very noticeable if it wasn't for the snow. It would be nice to see it in sunshine!
 
I guess, the last picture looks like more or less normal for this lenses. Possibly, it could be tuned a bit better, but not much. Also, I would suggest flashing my firmware mod for Git2P 90 (you can find it on this forum, and make sure official firmware ver. 1.2 is flashed before), selecting video mode 1080p30 (it has much better resolution compared to official one, as full readout mode is used), Sharpness Low (artificial sharpening is completely disabled) and Zoom to x1.6, and connect TV via micro HDMI. Such settings should help to adjust focus more precisely. Place focusing target in ~3-4 meters, and another one (something like book with pictures) in ~1.2-1.5m. Turn lenses in small steps until you get center and edges in focus and additionally check target which is near by, to make sure it is still more or less in focus but on the edge of DOF. In such way should get hyperfocal distance focusing, so everything in >~1 meter should be sharp.
 
I also think that last photo looks just about right. It only takes very small adjustments to go from in-focus to out-of-focus, especially with a longer focal length like this. Sometimes it takes a few goes to get it right. I would strongly suggest a locking collar to keep it in place if you're only relying on glue at the moment.

As Nigel says, try a few other subjects to test the focus. A large brick wall parallel to the sensor is good for checking detail across the frame.
 
A short interposed question of mine (sorry):
I would like to experiment with a 70° lens (or even less) to achieve a kind of zoom. Lots of different ones can be found from chinese dealers. Is there a way to tell good lenses from bad ones? Or anything else that I should care about?
 
A short interposed question of mine (sorry):
I would like to experiment with a 70° lens (or even less) to achieve a kind of zoom. Lots of different ones can be found from chinese dealers. Is there a way to tell good lenses from bad ones? Or anything else that I should care about?

Buy from a reputable seller. Many of us have had good experiences with Treeye on AliExpress. There are a few other AliExpress sellers worth looking into. From my experience their lenses have ranged from pretty good to excellent in quality and Treeye is a friendly, responsive and honest seller. Another option is to check out a specialty supplier like Peau Productions but be aware that they can be fairly pricey.

The thread Mobius Telephoto Dashcam, although not about the Git2 may provide some interesting and useful information if you are interested in a narrower FOV lens for your camera.
 
Last edited:
A short interposed question of mine (sorry):
I would like to experiment with a 70° lens (or even less) to achieve a kind of zoom. Lots of different ones can be found from chinese dealers. Is there a way to tell good lenses from bad ones? Or anything else that I should care about?
Ideally get one that is specified to work with your sensor, otherwise it will be a bit of a gamble unless you go on a recommendation. Also remember that the image stabilisation works correctly with the FOV it is calibrated for, if you have a lot of zoom then the stabilisation may make things worse rather than better - not an issue if used as a dashcam since you don't need stabilisation, but it is a big issue for some uses, and at 70 degrees without stabilisation it will be much more difficult to hold it steady than at wider FOVs.
 
@Nigel Stabilization will be done by a gimbal. That way the image will not be cropped, and I get the real resolution instead of an artificially inflated picture.
 
@drygdryg I have 2 lenses from Fulekan on Aliexpress. These lenses are 3MP only, but relatively fast and have low distortion. First one is 3mm f/2 for 1/2.7'' sensor, and I use it with Git1, and another one is 4.2mm f/1.8 and has been used with Git2P. Both lenses are big and heavy and won't fit into camera as original one, so some modding was required, and both have good coating, seems to be much better than one used in original lenses, with good contrast, colors and backlight resistance.
Also, @nutsey has reviewed their 12MP 3.2mm f/2 lenses and seems to be impressed with its resolution, but that one is really huge.
They have another lens - 12MP 8mm f/2, which have mentioned specs, if you do not have size/weight/distortion constraints. I guess, it should be quite good too.
 
Guys, thank you for your replies. But what I still don't understand is the resolution of a lens. I mean, a lens is 'moving' photons, there are no pixels inside the glass. The wavelength of light should be the lens resolution, not pixels. Does one of you know the secret?
 
Guys, thank you for your replies. But what I still don't understand is the resolution of a lens. I mean, a lens is 'moving' photons, there are no pixels inside the glass. The wavelength of light should be the lens resolution, not pixels. Does one of you know the secret?
If you use a 16MP sensor, but only a 3MP lens then the image will only show 3MP of detail. Some lenses can produce sharper images than others for various reasons, probably the accuracy of the moulding/polishing is the main one. A high resolution sensor can't make a blurry image look sharp.
 
Place focusing target in ~3-4 meters, and another one (something like book with pictures) in ~1.2-1.5m. Turn lenses in small steps until you get center and edges in focus and additionally check target which is near by, to make sure it is still more or less in focus but on the edge of DOF. In such way should get hyperfocal distance focusing, so everything in >~1 meter should be sharp.

Hi guys, I just wanted to update you on the situation with the lens.
As per above instructions I managed today to set up a test scene (indoors unfortunately) and took a couple of sample pics to check the focus.
I set the camera to ISO 50, stabilised it and set it on a 3 sec timer for good measure. I used the modded 90 FOV firmware and set sharpness and contrast to normal. Interestingly although I set the ISO to 50, the camera overrode that and used ISO160.

The scene was photographed from 1m away and then from 1.5m away. I think that the results were rather acceptable.
(To my eyes there is a noticeable softness to the objects due to the focus but I could be wrong and I'm simply asking for a bit too much from a fixed focus camera?)

1m away

1.5m away

The actual usage of this camera is for video recording whilst riding my motorcycle. So in practice most objects would -hopefully! :p - be further away than 1-1.5m. What do you think? Should I leave it at that or would you tinker with the lens more?

On another note, do you think GitUp could sell me a 90 FOV front cover to properly match the longer lens?
 
I used the modded 90 FOV firmware and set sharpness and contrast to normal. Interestingly although I set the ISO to 50, the camera overrode that and used ISO160.
Original firmware actually uses ISO Auto for ISO50 and ISO100, if shutter speed set to Auto. Modded firmware has been added ISO100 support (instead of ISO1600) for Shutter speed Auto , but ISO50 is still not working as it should. So, I would suggest stick to ISO100 in such cases, or set shutter speed manually too, if ISO50 or ISO1600 is required for stills.
Focus seems to be ok for me. Also, please note, that modded firmware has reduced sharpening settings compared to original one to make image more natural and reduce noise introduced by edge enhancement , so it will have noticeably softer look too.
 
Hi guys, I just wanted to update you on the situation with the lens.
As per above instructions I managed today to set up a test scene (indoors unfortunately) and took a couple of sample pics to check the focus.
I set the camera to ISO 50, stabilised it and set it on a 3 sec timer for good measure. I used the modded 90 FOV firmware and set sharpness and contrast to normal. Interestingly although I set the ISO to 50, the camera overrode that and used ISO160.

The scene was photographed from 1m away and then from 1.5m away. I think that the results were rather acceptable.
(To my eyes there is a noticeable softness to the objects due to the focus but I could be wrong and I'm simply asking for a bit too much from a fixed focus camera?)

1m away

1.5m away

The actual usage of this camera is for video recording whilst riding my motorcycle. So in practice most objects would -hopefully! :p - be further away than 1-1.5m. What do you think? Should I leave it at that or would you tinker with the lens more?

On another note, do you think GitUp could sell me a 90 FOV front cover to properly match the longer lens?
I think that to use manual ISO, you also need to set a manual shutter speed. Doesn't match the above post, but that is how I always understood it on the Sony version.

For a 90 degree lens, 1m to infinity is a bit of a challenge, it doesn't really have enough depth of field. For motorcycle use I would be trying to focus it for 2m to infinity which should be much easier. For the 170 degree lens it should do 0.8m to infinity.

You can ask Gitup for a new front, they are often helpful. I often have a 52mm filter holder on mine which also solves the problem, and if I don't have a filter on it I still use it for a 52mm lens cap which I prefer to the little silicone one.

I prefer to add a little sharpening in the editor rather than in the camera, if I'm editing it anyway then it ends up with a better result - better sharpening and better colours.
 
I think that to use manual ISO, you also need to set a manual shutter speed.
This is true for video mode in original firmware, but setting ISO with Shutter Speed Auto kind of works in photo mode, but exception is ISO50 and ISO100. Also, it ignores ISO settings with Shutter Auto if it is too dark, so ISO3200 will be used, while exif shows ISO set. Such behavior has been changed in modded firmware - ISO setting in video mode specify max ISO (ISO limit) to be used (except ISO50, which works as Auto), and photo mode should respect ISO settings (ISO 100-800), without falling back to ISO3200.
 
Going into the technical nitty gritty I have a question regarding the lens' focal lengths and respective FOVs.

The sensor used in the Git2P is a 1/2.33 type which measures exactly 6.18372 mm x 4.65648 mm.
This means a crop factor of 5.6, so a 3.87 mm lens gives a FOV of exactly 90 degrees or roughly 22 mm focal length in 35 mm equivalent.
Interestingly enough, the 4.1 mm lens has a FOV of 86.7 degrees or 23 mm equiv focal length.

Now the question is this: with the sensor size being a given, what focal length (actual/35 mm equiv) is the 170 FOV lens?
According to this calculator it would have to be a ridiculously small length; 0.35 mm actual / 1.96 mm equivalent!
Can this be? I mean, I know it's a fisheye and all but I've never seen such specs on lenses before.
 
Thread starter Similar threads Forum Replies Date
Sandro Git2 11
Back
Top