01304 827609 info@use-ip.co.uk Find us

HikVision ColorVu DS-2CD2347G1-LU - What's your favorite Image settings?

E-Q

New Member
Messages
2
Points
3
Hi Guys,

I just registered to say Thank You!
I was able to resolve several questions I had about the DS-2CD2347G1-LU ColorVu camera and this forum was the only one that was really helpful to me.

As I'm getting used to the camera, I would love to hear from you about your favorite Image settings.

I'm currently working on installing the camera to perform 24/7 surveillance of my driveway. I'm interested on a good setup for night time footage/recording.

Thanks in Advance!
 
Hikvision PTZ cameras
I prefer to disable the LED light on mine and rely on some low power ambient LED light fittings that I've installed. For nighttime I need to use the WDR to lighten the shadows as I have a white car directly in front of the camera which causes the camera to under expose the rest of the image. This causes a problem as there is a bug in the firmware (5.6.5).....

The WDR setting can be set to auto or manual and in both modes it can be set from 1 - 100. In manual mode this works correctly with level 1 providing minimal effect and 100 maximum. The auto setting, however doesn't appear to work - if you set it Auto and level 1, it activates WDR at a level roughly equivalent to 50 on the manual setting. In practice this means I have to use manual level 15 to get a correctly exposed night time image, which leaves the daytime image looking over exposed and washed out. Time slots can be used to switch the image settings but as has been mentioned numerous times by others, time slots are a hopeless method of altering the image settings as sunset time varies between 4PM and 10PM throughout the year altering by up to 4 minutes per day. The auto setting if it worked correctly would be useful. It should apply WDR automatically, according to the image exposure, ramping the WDR level up to the value at which the user has capped it, therefore solving the issue caused by changing lighting conditions.

For most of the cameras I install, regardless of model, I have a few settings that I start with:
  1. Contrast - I usually lower this and have found that on most of the current camera range it makes a big difference to the night time images. At default the shadows seem to have no detail. Lowering it from 50 to about 35 greatly improves nighttime images, evening the exposure, without adversely affecting the daytime image. Note that those values are in the web configuration - the NVR local monitor shows different values (39 default rather than 50)
  2. WDR - see above. I set this no higher than 15.
  3. I Frame - Hikvision set all of their camera I Frame default values at 50. As they also default all their cameras to 25 FPS, I maintain this 2 second I Frame interval. If I install a camera set to 10 FPS - I log into the camera and set the I Frame to 20 for example. Only available in the cameras web GUI.
  4. Smoothing - In the cameras web GUI. Level 1 (clear) to 100 (smooth) is set to 50 by default. I tend to drag this all the way down to 1.
  5. Shutter speed - Hikvision default all their cameras to 1/25 sec which is too slow. ColorVu are set even slower by default at 1/12 sec. It may look fine live, but when you playback and try to get a still shot of the local scrote breaking into your car, you'll see too much motion blur to resolve detail. I set this to a minimum of 1/50 sec but prefer 1/100 sec or higher. Unfortunately this is a huge tradeoff as each time you double the shutter speed, you halve the sensors available light.
Some time ago on a Hikvision training course, the instructor brought up a night time image from a "Powered by Darkfighter" camera saying "this what you can achieve when set up correctly" All he'd done was set a ridiculously low shutter speed, and set the day to night switch at 7 to keep it in colour mode. The image was horrific, completely unusable and any movement in the image would have just been a smudge on the screen. It's as well it was a webinar not a live training session as I laughed out loud (while some of other attendees excitedly asked what model it was and how to set it that way)
 
Last edited:
I prefer to disable the LED light on mine and rely on some low power ambient LED light fittings that I've installed. For nighttime I need to use the WDR to lighten the shadows as I have a white car directly in front of the camera which causes the camera to under expose the rest of the image. This causes a problem as there is a bug in the firmware (5.6.5).....

The WDR setting can be set to auto or manual and in both modes it can be set from 1 - 100. In manual mode this works correctly with level 1 providing minimal effect and 100 maximum. The auto setting, however doesn't appear to work - if you set it Auto and level 1, it activates WDR at a level roughly equivalent to 50 on the manual setting. In practice this means I have to use manual level 15 to get a correctly exposed night time image, which leaves the daytime image looking over exposed and washed out. Time slots can be used to switch the image settings but as has been mentioned numerous times by others, time slots are a hopeless method of altering the image settings as sunset time varies between 4PM and 10PM throughout the year altering by up to 4 minutes per day. The auto setting if it worked correctly would be useful. It should apply WDR automatically, according to the image exposure, ramping the WDR level up to the value at which the user has capped it, therefore solving the issue caused by changing lighting conditions.

For most of the cameras I install, regardless of model, I have a few settings that I start with:
  1. Contrast - I usually lower this and have found that on most of the current camera range it makes a big difference to the night time images. At default the shadows seem to have no detail. Lowering it from 50 to about 35 greatly improves nighttime images, evening the exposure, without adversely affecting the daytime image. Note that those values are in the web configuration - the NVR local monitor shows different values (39 default rather than 50)
  2. WDR - see above. I set this no higher than 15.
  3. I Frame - Hikvision set all of their camera I Frame default values at 50. As they also default all their cameras to 25 FPS, I maintain this 2 second I Frame interval. If I install a camera set to 10 FPS - I log into the camera and set the I Frame to 20 for example. Only available in the cameras web GUI.
  4. Smoothing - In the cameras web GUI. Level 1 (clear) to 100 (smooth) is set to 50 by default. I tend to drag this all the way down to 1.
  5. Shutter speed - Hikvision default all their cameras to 1/25 sec which is too slow. ColorVu are set even slower by default at 1/12 sec. It may look fine live, but when you playback and try to get a still shot of the local scrote breaking into your car, you'll see too much motion blur to resolve detail. I set this to a minimum of 1/50 sec but prefer 1/100 sec or higher. Unfortunately this is a huge tradeoff as each time you double the shutter speed, you halve the sensors available light.
Some time ago on a Hikvision training course, the instructor brought up a night time image from a "Powered by Darkfighter" camera saying "this what you can achieve when set up correctly" All he'd done was set a ridiculously low shutter speed, and set the day to night switch at 7 to keep it in colour mode. The image was horrific, completely unusable and any movement in the image would have just been a smudge on the screen. It's as well it was a webinar not a live training session as I laughed out loud (while some of other attendees excitedly asked what model it was and how to set it that way)

This is amazing -Thank You so much!
 
I prefer to disable the LED light on mine and rely on some low power ambient LED light fittings that I've installed. For nighttime I need to use the WDR to lighten the shadows as I have a white car directly in front of the camera which causes the camera to under expose the rest of the image. This causes a problem as there is a bug in the firmware (5.6.5).....

The WDR setting can be set to auto or manual and in both modes it can be set from 1 - 100. In manual mode this works correctly with level 1 providing minimal effect and 100 maximum. The auto setting, however doesn't appear to work - if you set it Auto and level 1, it activates WDR at a level roughly equivalent to 50 on the manual setting. In practice this means I have to use manual level 15 to get a correctly exposed night time image, which leaves the daytime image looking over exposed and washed out. Time slots can be used to switch the image settings but as has been mentioned numerous times by others, time slots are a hopeless method of altering the image settings as sunset time varies between 4PM and 10PM throughout the year altering by up to 4 minutes per day. The auto setting if it worked correctly would be useful. It should apply WDR automatically, according to the image exposure, ramping the WDR level up to the value at which the user has capped it, therefore solving the issue caused by changing lighting conditions.

For most of the cameras I install, regardless of model, I have a few settings that I start with:
  1. Contrast - I usually lower this and have found that on most of the current camera range it makes a big difference to the night time images. At default the shadows seem to have no detail. Lowering it from 50 to about 35 greatly improves nighttime images, evening the exposure, without adversely affecting the daytime image. Note that those values are in the web configuration - the NVR local monitor shows different values (39 default rather than 50)
  2. WDR - see above. I set this no higher than 15.
  3. I Frame - Hikvision set all of their camera I Frame default values at 50. As they also default all their cameras to 25 FPS, I maintain this 2 second I Frame interval. If I install a camera set to 10 FPS - I log into the camera and set the I Frame to 20 for example. Only available in the cameras web GUI.
  4. Smoothing - In the cameras web GUI. Level 1 (clear) to 100 (smooth) is set to 50 by default. I tend to drag this all the way down to 1.
  5. Shutter speed - Hikvision default all their cameras to 1/25 sec which is too slow. ColorVu are set even slower by default at 1/12 sec. It may look fine live, but when you playback and try to get a still shot of the local scrote breaking into your car, you'll see too much motion blur to resolve detail. I set this to a minimum of 1/50 sec but prefer 1/100 sec or higher. Unfortunately this is a huge tradeoff as each time you double the shutter speed, you halve the sensors available light.
Some time ago on a Hikvision training course, the instructor brought up a night time image from a "Powered by Darkfighter" camera saying "this what you can achieve when set up correctly" All he'd done was set a ridiculously low shutter speed, and set the day to night switch at 7 to keep it in colour mode. The image was horrific, completely unusable and any movement in the image would have just been a smudge on the screen. It's as well it was a webinar not a live training session as I laughed out loud (while some of other attendees excitedly asked what model it was and how to set it that way)

Really helpful post as I am also playing with settings.

Quick question - Do you prefer H264 over the other variants in terms of best image quality?

Thanks again Dan
 
Really helpful post as I am also playing with settings.

Quick question - Do you prefer H264 over the other variants in terms of best image quality?

Thanks again Dan

Thanks Dan. To be honest I was slow to switch over to H265 when it first became an available option in the Hikvision range a few years ago but now it’s all I use. I won’t use H265+ as when I have I’ve seen all sorts of unwanted artefacts - ghost trails, repeat images etc. A while ago, believing that the H265 algorithm couldn’t be saving so much space without discarding viable image details, I did a little trial. For a couple of nights I had two identical models of camera, with the same view, available light etc and set to 25FPS with bitrates set to Hiks recommended (for the respective codec) plus about 50 percent - one on H264 and one H265. With my eyes I couldn’t see a discernible difference when comparing side by side.
 
That’s great info - has given me the confidence to go with 265!

one more ask if that’s ok. Current camera settings I have are in the pic . Any changes you would recommend for best video quality (particularly at night)

37E3612E-6B17-46D0-9670-280F15DCF813.jpeg
 
They look fine - Hikvision specify the bit rate at 25FPS as 4096 for that resolution on H265 so at 6144 you may be able to set the quality to Highest as opposed to Higher though whether you’ll see the difference is another matter.

For nighttime image quality (or anytime for that matter), the important settings are the Image settings where you can adjust WDR, Contrast, Shutter speed etc. It’s best to work on one setting at a time to see the effect and ensure the change doesn’t adversely affect the daytime images.
 
  • Like
Reactions: E-Q
Awesome - thanks again. Bitrate is the one that I know least about.

Am having fun with the image settings plus adding my own exterior lighting. Your advice has massively helped on this!
 
Good afternoon, I'm terribly sorry for hijacking this particular thread, this is such an incredibly helpful post and has already helped provide some invaluable settings info for my cameras. JB1970, could you possibly spare a moment to have a look at my settings for my two DS-2CD2386G2-IU? Many thanks in advance. Richard
 
Good afternoon, I'm terribly sorry for hijacking this particular thread, this is such an incredibly helpful post and has already helped provide some invaluable settings info for my cameras. JB1970, could you possibly spare a moment to have a look at my settings for my two DS-2CD2386G2-IU? Many thanks in advance. Richard
Feel free to post your settings. Do keep in mind that there are no magic settings. Every scene and environment is different with regard to the viewing angle, area of interest, available lighting etc. A lot just needs to be tweaked by eye to give a nighttime image that is a compromise between being bright enough and clean enough.
 
I hope this does not sound like I'm being crazy here, but here goes ;)
What with all the current AI and Machine Learning being built into so many products these days.
Would is not be to much to expect these camera's to be able to self adjust based upon the current scene/weather conditions/time of day
As opposed to the user needing to set settings, and select options which are fixed?

It struck me today actually here in the UK where it snowed for the 1st time.
I have the following:

Hikvision 8mp 4K Turret
Reolink 8mp 4K Turret
Nest Hello DoorBell camera.

Can you guess which one of these three were able to handle the sudden white everywhere scene that happened this morning?

It was the Nest Hello Doorbell !

Both the HikVision and Reolink seemed to be so overwhelmed by the amount of white both on the ground and in the sky that nearby tree's and hedges were very dark. whereas the Nest doorbell seemed more able to balance this out, and maintain exposure compensating for it all.

One could perhaps hope that the makers of these CCTV camera's would set-up a built in knowledge of many scene's and scenarios, as I understand modern mobile phones do, so that they know how to handle and adjust automatically how they deal with various conditions, and not have to have a user alter fixed settings which may well only be appropriate for one given condition.

Would this be a reasonable expectation?
 
I hope this does not sound like I'm being crazy here, but here goes ;)
What with all the current AI and Machine Learning being built into so many products these days.
Would is not be to much to expect these camera's to be able to self adjust based upon the current scene/weather conditions/time of day
As opposed to the user needing to set settings, and select options which are fixed?

It struck me today actually here in the UK where it snowed for the 1st time.
I have the following:

Hikvision 8mp 4K Turret
Reolink 8mp 4K Turret
Nest Hello DoorBell camera.

Can you guess which one of these three were able to handle the sudden white everywhere scene that happened this morning?

It was the Nest Hello Doorbell !

Both the HikVision and Reolink seemed to be so overwhelmed by the amount of white both on the ground and in the sky that nearby tree's and hedges were very dark. whereas the Nest doorbell seemed more able to balance this out, and maintain exposure compensating for it all.

One could perhaps hope that the makers of these CCTV camera's would set-up a built in knowledge of many scene's and scenarios, as I understand modern mobile phones do, so that they know how to handle and adjust automatically how they deal with various conditions, and not have to have a user alter fixed settings which may well only be appropriate for one given condition.

Would this be a reasonable expectation?
I'm not surprised at all by what you've found. The Nest is a consumer product designed to be installed by anyone, whereas the Hikvision is a professional product. It's akin to taking a snap with the latest iPhone/Samsung smartphone and taking the same image with a full frame mirrorless or SLR camera using RAW mode. Without the knowledge of what camera settings to use pre capture and what image processing to apply post capture, the smartphone image will provide a far more pleasing picture. That's not to say it's a better camera - it's not, it's a phone, but with any professional product, exposure and image settings to some extent are left to the installer to match the scene.
 
I'm not surprised at all by what you've found. The Nest is a consumer product designed to be installed by anyone, whereas the Hikvision is a professional product. It's akin to taking a snap with the latest iPhone/Samsung smartphone and taking the same image with a full frame mirrorless or SLR camera using RAW mode. Without the knowledge of what camera settings to use pre capture and what image processing to apply post capture, the smartphone image will provide a far more pleasing picture. That's not to say it's a better camera - it's not, it's a phone, but with any professional product, exposure and image settings to some extent are left to the installer to match the scene.

Thanks for the interesting reply, and I don't disagree with the fact that they are perhaps aimed at different sectors.
I'm not sure I'm in full agreement that, AI and Deep learning is something only suitable for a home user.

However, perhaps I did not correctly explain my point.
You correctly mentioned that you'd wish the professional installer (and let's assume for now that the installer is a good one who actually knows their job perfectly) to set the image up for you.
That's great.

But what do they set the many manual image setting up perfectly for?
The super bright sunny day?
The overcast autumn day?
The blinding white everywhere of a winters snow day?
The dark at night allyway?
The same alley way when the lights come on?

Hence why I'm suggesting in 2020/2021 perhaps these camera's can take a book from the GIANT's and incorporate 'models' into their systems to adjust for different scenes.

I know people like Samsung etc etc have thousands of potential scenes set up, sun behind you, sun in front of you, lighting from various directions, people, pet's, landscapes, weather conditions, and will alter settings to match what's being seen by the sensor right now for the best output.

Think of it, like a trained installer being there live, all the time adjusting your camera all the time as the scene it's looking at changes.
And we can't have that can we :)

But we can have it, when it's built in via learning, and pre-taught algorithms.

You'd not set your SLR up for a perfect scene, then walk away and leave it with those settings now would you ;)
 
Thanks for the interesting reply, and I don't disagree with the fact that they are perhaps aimed at different sectors.
I'm not sure I'm in full agreement that, AI and Deep learning is something only suitable for a home user.
Agreed. But I wouldn't want the camera in total control of the settings. AI and machine learning cannot read my mind as to how I want an image and on which element of the scene is the priority with regard to exposure.

You correctly mentioned that you'd wish the professional installer (and let's assume for now that the installer is a good one who actually knows their job perfectly) to set the image up for you.
30 odd years - not perfect but I'm pretty good ;)

But what do they set the many manual image setting up perfectly for?
The super bright sunny day?
The overcast autumn day?
The blinding white everywhere of a winters snow day?
The dark at night allyway?
The same alley way when the lights come on?
Some of this can be accounted for with camera positioning, scheduled settings etc but there will always be some compromise.

Hikvision can't yet schedule an image change based on an offset to sunrise/sunset or perfect the human detection algorithms to differentiate between humans and animals so you may be waiting a long time.
 
Thanks for the reply and I may question if you'd like what AI did or did not do, until we see it :)

And indeed, CCTV does seem to lag a lot behind what we are all used to in our (for example) Phones from even 5 years ago.
Most modern phones, will destroy a security camera for image quality.
Higher pixel count of course but also the tech that build into the "Chippery" that handles the data the sensor is collecting.

Took a shot with my new 8MP Hikvision and my 3 generation old Google Pixel phone and, well, pretty embarrassing.

Apologies if I'm sounding a bit negative. All I really want it to see this improve, and fear there is little to no reason for them to put much time/money into it, as it's "good enough"
All I'd like to see is enough of a reason to give them enough of a kick up the backside to improve.

I was able to go through my photo library and find an example.
Slightly different angle, but pretty much the same distance.
New Hikvision 8MP Turret, vs my Pixel 3 phone Taken at the same time (same shadows) :)


1040724d05e74ef3346100c9f196961d.jpg


acffb28f6e0fc5dec408d2571ee6bd97.jpg
 
It would appear from the image that you've maybe taken a snapshot on the Hikvision camera using Hik Connect? and may have used the sub stream rather than the main stream. We've no idea what compression you're using or any of the settings on that Hikvision camera. You may also have digitally zoomed or cropped that image as the FOV appears very narrow. Regardless of all of that, you're comparing photo to video - JPEG vs H264/265/265+ and the associated compression (the pic from the Hik is unlikely to be the I frame) For the sake of argument, lets say the Hikvision 8MP camera is streaming at 20 frames per second utilising 10 Mbps (as mine are). A still from that main stream is just 941KB.

Things have improved monumentally over the last 8 years or so and you actually get a lot for your money with Hikvision quality wise. I totted up the cost of my previous 4 camera analogue CCTV system. 10 years ago it would have been £2,700 plus VAT, trade cost, materials only (maybe £4K installed). For that I got a DVR with 1 TB hard drive and some quality 1/3" Panasonic cameras with resolution of 600 TV lines.

There are better cameras available at a higher price - you get what you pay for.

I'll leave it at that as the topic of the thread has drifted somewhat.
 
Last edited:
Yeah, it's a fair point, and I'll admit it's not probably a fair comparison, as you say this is a snapshot of a video frame as opposed to a static photo.
Perhaps one day the two can become a bit closer :)
For your info, this was taken as the best quality settings I could get from the HikVision and it's zoomed in to 8X to get close to the cereal box, as was the phone image zoomed in also.
For general reference, (in case anyone is interested)
The 1st image in the above post, as I said was from a new 8MP 4K Hikvision Turret
The image below is the same thing from a new 8MP 4K Reolink Turret.
Slightly different angle, but same distance and zoomed in the same.
If nothing else it's interesting to compare detail/quality with the same scene at the same time:
Unsure which would be considered best, they both appear to have good and bad points:

a8c6d6f529fbc41a0758fa66c1538ad6.jpg
 
The phone will likely have had some native zoom/optical zoom so it wouldn't degrade the image quality so much (up to a point). "Zooming in" on a fixed lens CCTV camera (cropping) can also reveal more detail, but only when you're zooming to match the two resolutions. IE if you are viewing a 4096 x 2160 (4K) camera on a 1920 x 1080P (full HD monitor/TV) then you can use the digital zoom a little, but once the number of pixels represented on screen is lower than the screen resolution you're just degrading the image.

Your best bet for comparison is to use a web browser, make sure you're viewing the main stream and export a snapshot from the cameras. View that exported image using any picture viewer but ensure you set the view to 100% or Actual Size, rather than fit to screen. This would allow you to evaluate the detail in each. The cameras would have to have a similar viewing angle for a fair comparison.

I mentioned "you get what you pay for". My aim has always been for the best performance in low light. I always use this as an example - search YouTube for the Sony VB770. That shows you what is possible with very deep pockets!
 
If phones were a good cctv option then samsung and Apple would be marketing the fact to bits. They don't and they are not, especially in low light. My EOS 80d and 70 to 300 L lens still out performs my samsung S20. Consumer phone cameras are just that. Built for consumers.

JB

Appreciate the info on settings. Agree totally, shutter speed V aperture size is one thing to get right for night shots.
 
I'm not surprised at all by what you've found. The Nest is a consumer product designed to be installed by anyone, whereas the Hikvision is a professional product. It's akin to taking a snap with the latest iPhone/Samsung smartphone and taking the same image with a full frame mirrorless or SLR camera using RAW mode. Without the knowledge of what camera settings to use pre capture and what image processing to apply post capture, the smartphone image will provide a far more pleasing picture. That's not to say it's a better camera - it's not, it's a phone, but with any professional product, exposure and image settings to some extent are left to the installer to match the scene.

Hi I hope you are well on the g2 version of this camera with the latest firmware does the auto wdr now work
 
Back
Top