I have been looking forward to writing this post since my last visit over to the home of Amped, Trieste in Italy, a few weeks ago.
Whilst there, I saw some of these features in early development, and identified immediately how they are going to help us all. Remember, I am just like you – an analyst who uses Amped FIVE every day to interrogate, interpret and investigate video.
As well as a number of hidden refinements and minor bug fixes, there are some major advances… so, grab a coffee (a big one) and let’s go take a look!
When you load an MPEG based video into FIVE, you will now see a new filter setting in the Video Loader Window. Color Range has three options as can be seen below…
‘From File’ reads the Color Range flag held within the metadata of the file. ‘Full’ and ‘Limited’ allow you to override this flag and display the video with either range. This is fantastic…. but I can hear a few people saying, “what do they mean?”
MPEG video that is designed for standard TV playback does not output all 256 levels of luminance. The levels are set to between 16-235. They are limited!
Full range has all 0-255 levels, so there are more shades of grey.
What happens then, when you have some footage that has been recorded with full range but the file presents the limited flag to a decoder? You get less values than are actually available.
Take a look at these frames from an MPEG video. On the left is the ‘From File’ version, being displayed with a Limited Color Range, and on the right I have forced the ‘Full’ Color Range.
The video was shot in extreme low light with the following Filters used after loading the video..
Both Exposure Filters had the same Parameters. The difference was Forcing the Full Color Range at the Video Loader stage.
In the ‘Limited’ display you can see that I have saturated pixels at top end of the scale, Blue equals totally black. However, in the ‘Full’ version, I do not.
So, in your dark CCTV footage that has been encoded using MPEG technology and exported for TV playback… could there be more hidden information in the upper and lower ends of the color range? … The answer, probably, is YES!
The interface for the Curves filter has had an overhaul. The graphical representation has been improved allowing a clearer visualization of the values.
Remember that The Curves filter maps the input image values to the output image values according to the drawn curve: the x coordinate of the curve is mapped to the corresponding y value. Every RGB channel of the image is processed according to the respective color curve. If the Value curve is chosen, all the channels are also processed in the same way.
A handy improvement has been made to the Filter Categories, with Interlace now having its own section.
This places the four Interlacing filters into their own category.
You may notice the filter ‘Line Doubling’. This used to be called Correct Aspect Ratio but in order to avoid confusion, it has had a name change to better explain the filters purpose.
Tools (File Info)
Following on from the Color Range, we now display this Range information within the File Info Tool. But that’s not the end of the story… there’s more, a lot more!
You will now see that the Storage Aspect Ratio (SAR), the Sample Aspect Ratio (SAR – yes same acronym), The Display Aspect Ratio (DAR) and the Pixel Aspect Ratio (PAR) are all displayed. These relate to a filter I will be coming onto next and a dedicated post is being written which will help you better understand their meaning.
It is important though to recognize the difference between the Storage Aspect Ratio and the Sample Aspect Ratio.
The video is stored by using 352 x 288 square pixels. Width / Height = Aspect Ratio (1.222). That is the Storage Aspect Ratio.
This MPEG video also has a Sample Aspect Ratio. This facility was developed for the MPEG4 container and the information is stored within each frames metadata. It is defined in this example as 12:11. Using this data ONLY, we can restore the Analogue to Digital (A > D) conversion.
Width x 12 = 4224, Height x 11 = 3168
4224 / 3168 = Display Aspect Ratio (1.333, or 4:3)
How about a standard PAL resolution of 720 x 576. If it had a SAR of 16:15, how should this be displayed according to it data?
720 x 16 = 11520, 576 x 15 = 8640
11520 / 8640 = 1.333 or 4:3
Any player being able to read the SAR will display the image at 768×576.
Note: There are studies ongoing surrounding the various issues encountered with Analogue originated video and the methods used by DVR’s to complete the A > D Conversion. If things change surrounding the math or details used – We will let you know!
How about another example…
Notice the Color Range on the example above…. FULL. Why? It’s a MJPEG.
All of the details have been taken from Amped FIVE’s internal decoder.
Furthermore, after scanning the file, the File Info Tool is now able to list all the streams and their type. This is a great way of identifying the audio or subtitle track you didn’t know about!
Linked with this, there is now a summary tab under the main File Info Window….
Remember, this is easily found by clicking the File Info icon….
Under the EDIT Filter Category, we now have a filter for correcting the Display Aspect Ratio.
This is commonly required when an analogue signal, designed to be displayed with a 4:3 Aspect Ratio, has been digitized and is then decoded using square pixels, rather than non-square pixels.
Again, this will be the subject of a further post, linked with the File Info reporting of SAR, DAR and PAR but, for the time being, let’s look at this in action…
The video file has a width of 704 and a height of 576. I can see that it has been encoded deinterleaved. It has originated from an analogue source.
After interleaving, and then deinterlacing (Remember: Now in their own category), I need to correct the Display Aspect Ratio to 4:3 to account for the analogue source.
You can see from the Filter settings, you are able to adjust the aspect ratio manually and also decide which axis to retain, the width or the height.
Well, standards dictate that it is the width that gets adjusted to compensate for the non-square pixels, ensuring all vertical samples are retained.
There may be times, however, when it is acceptable to adjust the height and retain the width. Retaining the width may allow small areas of interest to be preserved that would otherwise be lost in any downsampling.
You also have the option of choosing your interpolation method.
To look at this in greater detail, take a look at these two posts…
A quick one…. Guess what it does?
Yes – that’s correct… It plays the video in reverse! Simple, but highly effective.
Manually adding some Date or Time information is a regular occurrence. Especially when you are using raw streams extracted from a proprietary container. This is another Filter that has had an overhaul, giving you much more flexibility in what is displayed.
The Timestamp tab is where you enter your parameters. It is the next tab, that decides how we want to format those details.
Each part of the information (The Date and Time), can be turned off or turned on through the use of Macros. In the first example I have simply chosen a standard format – ‘Time with Milliseconds’. As such (after a font and size change) it appears like this…
However, we can change the order and even the type of information displayed…. let’s say we need to display what day of the week this was. By Manually adding %A before the time, we can add in the day of the week.
It now looks like this…
There are many options available to you and they are all listed in the Filter Reference HTML document. You can go direct to this section by clicking on the Filter Help..
Or you can see all the filters by opening the Filter Reference Guide. It’s here if you have forgotten! (Simply ‘Ctrl + f’, and then type Timestamp).
Whenever selecting an area is needed, you are now able to fine-tune its location and size manually.
The selected region in the Filter settings is locked. This will only display and record your input parameters. I find it easiest to simply draw an area on the image with the mouse cursor and then fine tune the settings within the Selector Tool via the keyboard. This is great when you want to get your crop areas to an even size, or you have specific x,y coordinates to select.
Quite often, when a new filter is implemented, there will be new functionality added over successive updates. Most of these come as a result of user feedback.
The Mixer filter now has enhanced usability in how to mix the two chains.
Firstly, the output size can be configured, taking parameter information from any of the selected options.
The Output Mode now has various options also…
These changes make comparative analysis much easier to conduct and present.
Anyone that has been following my own blog posts at Spreadys.com over the years, will know that I regularly utilize the information stored within the MPEG streams to verify the location, type and movement of Macroblocks. This is very important when critical information appears within a picture that sits in the middle of a Group of Pictures (GOP).
Under the Verify category, we now have a new filter ‘Macroblocks’.
When this is selected on an MPEG video, the filter settings require user input to enter the parameters.
The output format is required as a new video will be created. All the usual options are there, with Mjpeg my preferred choice. The new video is created in order to utilize FIVE’s ability to Mix chains…. I’ll show you an example at the end.
Finally, we have the option for what you wish to visualize. Only one type of Motion Vectors may be selected to avoid confusion! You don’t want arrows everywhere and not know which ones relate to which.
In order to understand this better, take a look at this short animation… some people may have seen this before, as I use it in some training classes!
Now let’s look at this ball, with the Macroblocks and Forward predicted Motion Vectors visualised…
So how do we interpret this new information? To make it easier I have selected an entire GOP, plus the next I frame, and then added the frame number – then cropped and exported…
Frame 20 is an I frame, so all the macroblocks are newly encoded. In this mpeg4 (Part 2), there is only one single type of newly encoded macroblock. These are all visualized with a red color.
Frame 21 is a P frame. Although the frame itself is predicted, there can be macroblocks that are newly encoded. So, in this frame we have a single new macroblock. The macroblocks shown in green have either a change in luminance or a change in location. The arrows highlight their motion vector. The gray areas have been retained from the previous frame.
Let’s now have a look at this with a piece of stock H264…
Can you see what’s new, what’s updated, and what’s newly encoded?
By utilizing ‘Mixer’, it is now possible to view both the original and the Macroblocks version side by side, or as an overlay.
To understand this better, take a look at this dedicated blog post on Macroblocks…
Big Bad Bugs!
- Linked images for the reports were not being displayed when certain naming conventions were within the file path. This has now been fixed, specifically allowing the hash ‘#’ within the folder name.
- Screen Capture. Random crashes were occurring with the Screen Capture function and those users on HiDPI screens were finding that the selection area was not being recorded correctly. Both of these issues have been fixed.
- When some files were being rewrapped or transcoded with Convert DVR, they were placed within the Input directory rather than a user specified directory. This has now been fixed.
- When selecting a Filter Category, you should only see the Filters relating to that set. There was an annoying bug that highlighted other filters from other categories. This bug has been fixed.
- When using the dark color themes, highlighted filters sometimes went a little too highlighted! Bug fixed – you can read them now!
- Some of the error reporting and warning messages have been rewritten to make them more understandable. (Perhaps Microsoft could do them same? HaHa)
Phew… so much new stuff! You have probably finished your coffee by now so, what are you waiting for….?
If you have an active support plan you can update to the latest version of Amped FIVE by going into the menu “Help” > “Check for Updates Online”. If you need to renew your SMS plan please contact us or one of our authorized partners.