MJPEG vs. H.264by John Honovich, IPVM posted on Apr 17, 2009 About John Contact John
Recently, IQinVision releaed an article advocating benefits of MJPEG.
[Update Dec 2010: We conducted extensive testing comparing MJPEG and H.264. Read our Test Results of MJPEG vs H.264.]
While I found the article technically accurate, well written and worth reading, the nature of the application and its economics demand that MJPEG be almost always avoided. Since H.264 is hot right now, this is a popular claim to make. However, a discussion of this can help examine the economics and operational drivers driving this interest.
Jason's central claims are:
1. With moving cameras or images of high activity areas, MPEG4 and H.264 provide little bandwidth savings relative to MJPEG.
2. Proper network design requires factoring in worse case scenarios so you will need to dedicate the same amount of bandwidth whether or not you use MJPEG, MPEG4 or H.264.
3. MJPEG provides higher quality because of no intra-frame compression.
4. Unlike MJPEG, with MPEG-4 vendors deviate from standards, increasing potential integation costs.
My counterpoints are:
1. For most users, cameras usually have low or modest activity, translating into significant savings for MPEG-4 or H.264. Most cameras in the world are fixed. Most cameras have significant periods during the day when there is little or no motion (nights, weekends, etc.) Even within PTZs, PTZs are often left at a home position, or iterate over a series of pre-sets stopping for 5 - 10 seconds each.
2. Many, perhaps most organizations, do not set network bandwidth budgets for worst case scenarios. Sometimes, organizations don't want to pay the money for the extra capacity but sometimes it can't be done due to constraints of reutilizing existing infrastructure (very common in wireless networks). In other words, organizations generally trade-off infrequent pixelization for immediate cost savings. Maybe this is 'objectively' wrong but this is common.
2a. Jason does not discuss storage but storage is a HUGE economic driver in the move away from MJPEG. I have had a number of occasions where my DVR/NVR with a 1TB hard drive was only recording for 13 days. Why? I had forgot we recently integrated just a few megapixel cameras using MJPEG. Let's say we can save 1 Mb/s by switching from MJPEG to MPEG4. Over a two month period, for one camera, that is 650 GBs. It would cost you $300 to $600 to add that much storage for each MJPEG camera.
3. As for quality, the difference in quality is usually close enough that most customers are ok with it, especially for the savings.
4. The issue with deviation from standards is generally a one-time cost/problem that can be amortized by the manufacturer over many different customers. In the larger scheme of things, it's mainly a nuisance.
In sum, then, the economics of reducing network and storage costs are usually very significant budgetary and operational factors that drive purchasing decisions. With megapixel manufacturers starting to announce H.264 support, it will be interesting to see what IQinVision does.
Comment #1 by Jason Spielfogel posted on Apr 20, 2008
Comment #2 by John Honovich posted on Apr 20, 2008
Most Recent Industry Reports
Testing Axis' Top Low Light Camera Q1635 on Nov 23, 2015
Low light performance continues to improve, first driven by advances in image processing and now increasing number of 1/2" imagers in 1080p HD cameras. IPVM has recently tested new super low light...
Audio Analytics Aggression Tested on Nov 20, 2015
What if you could use your IP cameras to detect fights before they start? That is the goal of Louroe / Sound Intelligence with their recently released Aggression Detector audio analytics. Cl...
Pelco Optera 12MP Multi-Imager Tested on Nov 09, 2015
This summer, Pelco came out firing against Arecont, touting the superior performance of its new multi-imager line vs Arecont's. But is this really the case? We bought a Pelco Optera 180° multi...
IP Camera Bootup Shootout 2015 on Nov 04, 2015
IP cameras, like PCs, take some time to boot up. And just like PCs, the amount of time can vary greatly. Many people do not care but some people find it annoying. Perhaps more importantly, in surve...
Live From China on Nov 02, 2015
China's growing influence, if not dominance, of the global video surveillance market is unquestionable. To better understand this, IPVM has gone to China. Our first stop is CPSE, which claims ~100...
Network Cabling for Video Surveillance Guide on Oct 30, 2015
In this 14 page guide, we teach the fundamentals of network cabling for video surveillance networks, how they should be installed, and the differences in testing them for production networks. Spec...
Large Video Surveillance Systems Guide on Oct 29, 2015
This 14 page guide explains the key uses, design factors, and players in the large system surveillance market. A global group of 80 integrators responded, each offering insig...
Sony 20MP / 4K Camera Tested on Oct 26, 2015
For 18 month, Sony has been hyping 4K cameras, a year before they even announced a 4K network camera. Now, amidst intense competition and price pressure, Sony has released their long awaited 2...
ONVIF Screen Capture Tested on Oct 23, 2015
Recording a PC's screen to a VMS has several uses, but historically has required expensive dedicated encoders or specialized software for each VMS. Now, a new offering called Screen ONVIF has...
Milestone Arcus VMS Tested on Oct 21, 2015
For more than a decade, Milestone was a Windows only VMS. With the Internet shifting power away from Windows OSes, Milestone launched a new VMS, called Arcus, which can be embedded onto Linux ...