This update to Amped FIVE includes many new additions and increases in proprietary format decoding. A new filter for the analysis of HEVC/H265 video, new functionality to Video Loader, Video Writer, and Timestamp adjustment, along with many improvements to Link filters, Multiview, and Annotate.
Hi everyone. It’s time once again for an Amped FIVE update. The Amped Software factory has been very busy over the last few months with many amazing new developments coming to our applications. This is the first of several updates for Amped FIVE this year so let us head straight in and see what we have in store for you.
See the New Features in Action
Formats
Amped Engine is the powerhouse behind the extraction and decoding of proprietary video formats. With the development of new technologies to improve multimedia and timestamp detection, we have spent the last few months re-engineering the data extraction methods for several common proprietary formats.
Although some of the file extensions are listed here, remember that you may have a completely different format, that just simply uses the same file extension. The file extension itself does not necessarily mean it’s a supported format.
M4V, CTF, MJP, and EZVP have all seen significant improvements in video stream and time data extraction.
Finally, and of particular note, is the AVE container format. Our new extraction method encompasses several different codecs, audio, and the all-important timing data. This container supports stream tiling, where several individual video streams are tiled together to construct a larger-resolution video. Our new extraction method handles the stream naming of the tiles where necessary.
Several other new formats have been added along with smaller improvements to many others. Thank you to everyone who has submitted files for analysis and support through to the codec team. With your help, we learned that these proprietary formats exist. Developing the decoders helps everyone within the forensic video community.
Copy and Verify
One of the very first stages of your investigation may be the movement of files from one location to another. This could be from network storage or a device attached to your workstation. The movement of that data must be controlled to ensure integrity. That is the main role of the Copy and Verify utility. However, remember that it can also be used for simple hash value calculations when required.
The first change here has been a redesign of the buttons and parameters. We found during Amped FIVE Training that many people would instinctively select Operation and Source Type first. Therefore, it made sense to move them to the top.
Next, we have a new Source Type.
“Multiple Files” allows you to select specific files for moving and hashing.
Imagine a USB drive with 2TB of data, but perhaps you only require 4 files.
Copy and Verify could be first used to create a full drive hash list to detail all the files and their associated hash values.
From this update, you will now be able to run a second pass but only select the files you need to be forensically copied. You are not restricted by folder structure anymore.
Video Loader
Following on from using Copy and Verify, you will usually head to Video Loader. It is here where we have added new functionality for handling standard multistream video formats.
A standard multistream video format will use a container that conforms to a documented standard. MKV or AVI, for example. The multistream part details the cases where several standard video streams are within a single container.
When a format of this type is loaded the following message will appear.
Selecting Yes will load all the streams into separate chains.
Selecting No loads the file without stream separation and hands the management of the streams to the single Video Loader.
To do this, a new option has been added to the filter.
As you can see above, there is only a single Video Loader. Now, you are able to select a specific video stream from within the file.
The numbering will match the stream identifier within the container.
If you always want to load all streams as separate chains, the default loading method can be selected in Program Options.
If you wish to extract the streams from their container, you can use Convert DVR from within Video Loader. This will work regardless of your loading method.
Convert DVR
Convert DVR has also received a small, but powerful, addition.
There is now a new option under the Audio tab that enables users to replace audio gaps with silence during an audio transcode stage.
Remember: you can stream copy the video but transcode the audio stream if required by specific circumstances.
Audio “dropout” is quite common and often found near the start of a recording. A common challenge is when there is a conflict between the duration of an audio packet, compared to the Presentation Time Stamp (PTS) of the next audio packet.
To visualize a dropout, here is a graph displaying the Audio PTS start times and the delta to the next packet. They should all be fairly close together in size. It’s not hard to spot that there is something a little different with one of them.
There is only 40ms of audio in packet 9, but the next packet does not start until another 700ms later.
There is a gap.
If these gaps are not taken into consideration when muxing video and audio from one container to another, synchronization errors may occur.
One of the solutions is to transcode the audio, and whilst doing so use the original PTS values to inject silence when there is a mismatch between the packet duration.
We hope that this small change will help with some of the many AV synchronization challenges that you face with CCTV recordings.
Coding Tree Units
Following on from using Convert DVR, you may need to evaluate the video compression. For some time now you could analyze the Macroblocks inside H264/AVC. Well, there is a new kid on the block, and it doesn’t use blocks at all!
Quick Intro to Coding Tree Units
H265, otherwise known as High Efficiency Video Coding (HEVC), uses Coding Tree Units (CTU). Each CTU is a 64×64 tile of pixels, that can contain separate Coding Units (CU). These can be a single 64×64 unit, or be further subdivided in 32×32 pixels, 16×16 pixels, and 8×8 pixels.
The CUs can be encoded using Intra prediction. A frame that can only use Intra prediction is an I Frame. CUs can also be encoded using Inter prediction from a previous reference frame. Frames that can only reference previous frames are called P (Predicted) Frames.
Finally, CUs can also be encoded using Inter prediction by using previous and future reference frames. Frames that can use both are B (Bi-directional) Frames. In both P and B Frames, CUs with no residual data can be skipped.
Visualizing CUs and Their Type
Visualizing CUs is now easy with this new filter which is found under the Verify category. Those used to analyzing H264 Macroblocks should be able to quickly understand the visual presentation and legend.
Here we have a small part of a video frame with the new Coding Tree Units filter applied.
Highlighted near the top are 3 CTUs that each consist of a single 64×64 pixel CU. Moving in closer we can visualize others.
As well as other CU sizes, you may be able to see some that are partitioned vertically or horizontally. CTUs allow for block partitioning which enables many different variations and then higher flexibility in the compression of that specific part.
Using the legend tab, we can interpret the color overlay. The legend is divided into two parts, with the first relating to the Block Type. Next, there is the color range for the Quantization Parameter.
From the legend, we can establish that the frame is constructed using Intra-encoding and Inter prediction from a previous frame.
There are also some CUs that have been skipped as they have no change in their encoded state.
Visualizing Motion Vectors and QP values
In addition to the CU type, users can overlay the Motion Vector data associated with all units containing encoding changes.
Lastly, users can visualize the quantization parameter associated with each 16×16 pixel coding unit. H265/HEVC uses a quantization range between 0-51.
Interpreting video compression is a key stage in the evaluation of CCTV and video evidence. Questions surrounding the reliability of objects and motion at a pixel, block, or unit level are commonly asked. Only after the compression used to construct the image has been understood, can an objective answer be given.
The prevalence of higher resolution cameras covering larger fields of view within our communities brings increased responsibilities on video compression to deal with all that data.
HEVC/H265 is a very clever and complex method to overcome some of these huge data challenges. However, it is then even more important to consider the purpose of video compression: to remove as much data as possible whilst retaining an acceptable visual quality.
The purpose of a camera may be to capture an entire intersection. Therefore the quality of the recording may be acceptable and visually perceived as very good. However, drill down to the CTU’s and the information available may be limited due to the requirement to remove much of the high-frequency detail.
Evaluation of the compression is sometimes just as useful for explaining why a task is not possible or why a question cannot be answered.
Adjust Timestamp
Using the extracted time data from proprietary formats is made easy due to the open nature of our .time files. They are formatted as Tab Separated Values (TSV) so they load into any application that can read the TSV format, such as Notepad++ or MS Excel.
However, it is often necessary to adjust timestamp data due to the requirement for interpolating or refining. All of those functions are within the Adjust Timestamp filter.
Within the filter, you now have the ability to write a new .time file, containing all the adjusted values.
Here we have part of an original .time file. We can see that the time data for each second is repeated. This is how it was stored in the proprietary file.
Using the filter, we can refine the milliseconds using a time source, such as the original Presentation Time Stamp (PTS), or through Linear interpolation.
This new, frame precise, data can then be displayed on the video. It can also be used for internal analysis, such as Speed Estimation or event timing. The ability to now write this adjusted data enables further external analysis, such as utilizing the many data functions in MS Excel.
Here we have the same range as above, and you can see that every adjusted row is marked with a *.
Finally, if you place the Adjust Timestamp filter after a Range Selector, and then write the file, it will only refer to that range.
The word “adjusted”, along with the date and time, will be appended to the new .time file.
Resize 1:1
Amped FIVE is being used more and more during the analysis of fingerprints, footwear, and tool marks. To assist those departments, we have made some changes to Resize 1:1 that also affects Image Writer.
After setting the real-world distance using a visual scale within the image, you now have options on how you wish to output this change.
As before, there is the ability for the image to be resampled, which changes the dimensions of the image using the selected interpolation method. The new option retains the pixel count but will change the Output DPI. This information will then pass through to the saved image as useable metadata.
Global Stabilization
This filter has been updated with the ability to increase the canvas size from within the filter settings.
When stabilizing a video, you will retain the original dimensions. Therefore, there may be some parts that extend beyond the visual area. One of the most common objects that may be required to be retained is a pixel-embedded timestamp.
Extending the canvas after a stabilization process allows users to visualize all, or a larger part, of the original video.
Annotate
It would not be an update to Amped FIVE without new functionality in Annotate. All the new additions this time are in the Hide Tool.
The first is a new effect called “Blacken”.
Guess what it does?
Yes, it completely blackens an area you wish to hide. If there is any concern over the future ability to un-pixelate or de-blur, then this new effect should help.
The next is “Edge Feathering”.
Again, it’s not hard to work out what this does.
The level of feathering is controlled by the Effect Strength slider.
Lastly, we now can invert the hide, seen below using the Pixelation effect.
After selecting your Hide effect and parameters, simply invert by checking the box.
This makes many redaction tasks super fast!
Link Filters
In all linking filters, such as Timeline or Multiview, we have simplified the Input options and removed the individual internal Filter ID for that Video Loader.
These can be visualized if required by selecting to show them within Program Options.
Multiview
Let us stay with Multiview and check out a new tab allowing you to change the background color.
Matching the color with Frame Size, or other filters, gives much more flexibility over the final presentation.
As there were only 7 cameras in the Multiview example, but we required a 3×3 grid, the other spaces now seamlessly match the larger presentation frame color in Frame Size. Increasing the frame size is often helpful in providing extra space for further annotations such as Magnify, as seen above.
Rotate
We have made a few small changes to the user interface within the filter settings.
As you can see the buttons now display the rotational angle. This is to assist in the interpretation of the resulting value. Often perceived as an error, a mathematically positive rotation is in a counterclockwise direction.
Therefore, rotating 90 degrees clockwise is -90. (Authors note: yes, I get confused too).
Video Writer
The Video Writer filter has had an overhaul, which includes some reorganization of the functions, and then some new capabilities added.
The Parameters tab now only contains the basic, necessary settings.
All other settings are now in the Advanced tab, along with some new ones.
The New GOP Size Parameter
The first new setting is “GOP Size”.
GOP, standing for Group of Pictures, details the amount of frames allowed between I Frames. For example, a GOP of 10 could be 1 I (Intra) Frame, followed by 9 P (Predicted) Frames.
This new option gives you much higher flexibility in your encoding quality and resulting file size. The next setting adds another option for more finer control.
The New B Frame Parameter
Using B (Bidirectional) Frames allows the encoder to identify differences in the frame from the previous one, but to look ahead and also use those differences to form the new frame.
These new parameters both combine to allow such settings as an I Frame only H264, which would have a GOP of 1, through to a GOP of 50, and include B Frames.
Depending on the resolution, the number of frames, and the complexity of the footage will all impact the resulting file size at the selected quality. However, the new settings can be evaluated to aid in retaining the quality but at a reduced file size.
Let us simplify that by using the following example frame data sizes.
In an example 12 frame GOP, if the frames were made up with all newly encoded data, it would be very large in file size. However, it would not use any prediction in the construction of each frame.
If we allowed prediction with P Frames, the file size would be considerably reduced. In the real world, it is often far lower than 50%. If we finally added the ability to use B Frames, the file size would be reduced further.
When evaluating the new settings remember to assess and quantify visual changes using Video Mixer. Using the file size of the originally loaded video may give you a false metric, as what is being written originates from your filter chain and not from the original file.
Compare Original
A quick way of reviewing and/or presenting a before and after image or video is the Compare Original filter.
From this update, users will have the ability to change the order of the two images.
This is particularly helpful when using modes that allow a half-and-half view as you can now select which side to focus on.
Generate Report
When generating PDF reports, you now have the ability to add left and right footers to the document.
When first opened, the data will be copied from the project properties, but both new fields allow for manual changes.
Other Updates
Finally a few more minor changes.
- Advanced File Info – New warning message
- Range Selector – Improved handling
- Sequence Writer – Improved project loading
- Installer – Standardized installation across all Amped Software applications
The only thing left is bug-busting!
Bug Fixes
Thanks to all the users who have reported bugs. As always, if you come across something that is not expected, please reach out to us in Support so we can fix it.
- Copy and Verify: fixed a bug causing file order to change when loaded in History using the right-click menu.
- Advanced File Info: Frame Analysis: fixed a bug where PTS Duration doesn’t calculate correctly with multistream formats.
- Advanced File Info: Frame Analysis: fixed a bug to allow Frame Analysis and GOP analysis if previously aborted by the user.
- Advanced File Info: fixed a bug where the stream identifier was incorrectly presented during frame hashing.
- Convert Frame Rate: fixed a crash caused by particular circumstances during Timeline export.
- Installer: fixed a bug that caused customized assistant scripts to be removed.
- Perspective Aligner: fixed the bug that caused an error message to appear when using the filter.
- Player: fixed a bug where the time counter failed to refresh after changing filter chains.
- Remove Duplicates/I Frames Selector: fixed a bug with filter settings not fully shown.
- Video Writer: fixed a bug where encoding is using a limited color range, even when the input is full.
- Fixed a crash related to decoding audio.
- Fixed a bug causing audio playback to slow down after Video Writer.
- Fixed a bug causing a lack of audio in custom video export.
- Fixed a bug that didn’t allow deleting the files on Windows until FIVE was closed.
Don’t Delay – Update Today
If you have an active support plan you can update straight away by going into the menu About > Check for Updates within Amped FIVE. If you need to renew your SMS plan, please contact us or one of our authorized partners. And remember that you can always manage your license and requests from the Amped Support Portal.
