Review: Dahua DH-IPC-MFW7442K1-Z4-T40 - Dual Sensor (4MP + 4MP) Bullet Camera

Very nice video and a lot to digest. I really liked the 5442 comparisons.

One thing though, as we are really interested in how well a face is captured, your tests do not allow any visuals of the face. If you are not comfortable showing your face, maybe you could ask a friend that does not care about being on camera to do the walk throughs?
 
Another job well done by our Treasury man in action. Great video and it puts every other reviewer out there to shame by being professionally done!

Is there a reason why you use H264H as opposed to H264?

Thanks for the kind words @sebastiantombs, much appreciated.


H.264H provides the best encoding with least PQ compromise in my experience on codecs supported by most cams so far. More complex algorithms while still performing well even when pushing higher bitrate and doesn’t crush blacks, lose details, add noise or appear flat. Makes a big difference at night. Certainly would recommend this over H.265 in current cam implementations any day of the week.
 
You're depth of knowledge is far beyond anyone else that I've encountered. Heck, just syncing the different cameras is a major undertaking and you do it so smoothly and that provides a true, real, side by side comparison that no one else has even thought of.

Being a Blue Iris user I have to ask how H264 performs?
 
I haven't seen H.264H mentioned before... is that H.264 but selecting profile="High Profile" (on a Hikvision camera)?
 
  • Like
Reactions: djernie
You're depth of knowledge is far beyond anyone else that I've encountered. Heck, just syncing the different cameras is a major undertaking and you do it so smoothly and that provides a true, real, side by side comparison that no one else has even thought of.

Being a Blue Iris user I have to ask how H264 performs?

@sebastiantombs another great question on H.264.

Yes you can use Main Profile H.264 or even H.264B (Baseline Profile) too if you wish (personally would steer clear of baseline). Let me go into some more detail below in case helpful for others here. Warning I’ll go into more technical detail for a sec but will try to keep it general where I can :)

H.264 Profile Differences, Impact & Summary

The differences in a codec profile really come down to the algorithms and features offered by the profile itself (increasing as you move from baseline, main, high) within the codec (where applicable). For example with H.264, when you move up from Baseline to Main & High profiles you start to gain extra features such as B Slices, CABAC or Context-based Adaptive Binary Arithmetic Coding (also used in HEVC or H.265 too BTW) vs the less performant/less efficient Context Based Adaptive Variable Length or CAVLC coding. CABAC aims to improve compression quality by moving to an adaptive approach to probability distribution that plays a part when the codec is looking at predictive vs actual blocks for areas to maximize compression.​
The high profile also gains 8x8 vs 4x4 transform adaptivity which allows the codec to choose between 8x8 or 4x4 block sizes in processing the image vs 4x4 only in the Baseline profile. This assists in overall output as it allows the codec to identify stronger motion isolation where it can, therefore leading to a more balanced and higher quality image (less artifacts, impacts to PQ etc) vs Baseline while still doing a good job on compression overall. High Profile was originally primarily focused at broadcast based applications BUT as IP cams for example have moved into HD & above, they benefit from these improved profiles (and newer codecs) more now.​
Therefore H.264 Baseline for example can be summed up as using fewer ‘encoding tools’ or less complex algorithms that can result in lower quality output BUT easier to decode by a wider number of systems while decreasing computational complexity. H.264H adds the more complex algorithms (some of which I mentioned above) while improving quality and efficiency and while being more complex it is NOT as complex (or CPU intensive) as H.265 generally speaking.​

Other Areas of Concern To Consider When Selecting a Codec & Profile

Over and above what I mentioned, in terms of where this will impact, most people focus on file size and have the mindset that as you move up codecs such as to H.265 you gain efficiency which should = maximizing quality while minimizing file size. To be fair that can be (and often is) true BUT as I mentioned above there are other criteria you should consider. One of these is that H.265 and other codecs also cause an impact to CPU/processing. Now, depending on device and a decent HW encoder, those CPU impacts can be minimal BUT on other mid to low end systems OR using devices such as IP cams trying to process video in real-time (on-board encoding of bitstream etc) it can cause issues with cam performance. This can be seen in a number of ways including but not limited to slow down/impact to responsiveness in GUI (if accessing live or from multiple target points such as NVR, 3rd party stream processor, multiple sub streams etc), degraded performance in stream output (which can be seen in fps or other areas) etc. This performance degradation can be quite impactful especially to critical captures IF cams are underpowered and/or processing becomes more complex in conjunction with other camera functions operating in real-time as I’ve written about before.​
As I mentioned above though, you have to take into consideration the specific types of footage you are capturing such as high speed, dark scenes, complex patterns, contrasting areas etc. Choosing the wrong codec can lead to great efficiency at the expense of sub-optimal and/or non-consistent quality in part or all of your footage (depending on content within it as stated). It is this impact to output quality that is not a compromise I advocate for, storage is cheap :)

In my experience, the compression efficiency vs PQ of H.264 in the High Profile (H) is the sweet spot for these cams right now. Thats not to say this sweet spot won't change (and I'll certainly update when I see it) as cams, codecs and configs evolve but my recommendation right now would be H.264H.

While on the subject of encoding, the other area that I’ve mentioned before is CBR vs VBR, always use CBR. VBR still doesn’t respond quick enough to real-time changes in motion to produce guaranteed great PQ that you can rely on in my experience. Often times you’ll see macro blocking or other compression artifacts that can ruin that critical frame etc. For those using VBR, try CBR and see what you’re missing ;)

HTH
 
Last edited:
Thanks for the explanation. It might be time to experiment with the 264 codec family and see what happens with my use case.
 
All, just a quick update on the nighttime video. I've found some other areas I want to pull apart during nighttime testing I've already captured so will be setting up to grab additional footage this week. This will delay me editing and publishing the video a few days but my goal is to still get this complete by end of week. Stay tuned
 
All, looks like YouTube finally finished processing this 8 Gig file :) This is a long video, weighing in at around 21 minutes but I’ve tried to capture all aspects of testing with this camera and it’s dual sensors. I’ve also included detailed timestamps so you can jump around or back as you wish.

Definitely watch the video through as there are some surprises here, good and bad with regards to this camera. I will cover more on these in my summary / wrap up and will also do a video of the menu so you see options for each lens etc too as that is important in understanding what you can and can’t do with the unit.

I’ve made a list of items such as bugs, areas that need addressing and design issues with recommendations for Dahua

Posting video link here and on the main page for reference

 
That was a very complete testing scenario video. Thank you. I was especially interested in the last set, using the CH1 for LPR and the CH2 in color for info on the car.

I really hope that when the Z7 comes out that the CH2 cam will be in 6mm. Maybe @EMPIRETECANDY can place a bug in Dahua's ear about making the CH2 available with 3.8 or 6mm lenses.
 
All, looks like YouTube finally finished processing this 8 Gig file :) This is a long video, weighing in at around 21 minutes but I’ve tried to capture all aspects of testing with this camera and it’s dual sensors. I’ve also included detailed timestamps so you can jump around or back as you wish.

Definitely watch the video through as there are some surprises here, good and bad with regards to this camera. I will cover more on these in my summary / wrap up and will also do a video of the menu so you see options for each lens etc too as that is important in understanding what you can and can’t do with the unit.

I’ve made a list of items such as bugs, areas that need addressing and design issues with recommendations for Dahua

Posting video link here and on the main page for reference


For your testing of LPR, may I ask what the distance was? I'd be looking at around 100' for my application.
Thanks!
 
  • Like
Reactions: djernie