Recherche avancée

Sur d’autres sites (8900)

• FFmpeg avcodec_open2 throws -22 ("Invalid Argument")

  14 avril 2023, par stupidbutseeking

  I got stuck trying to write a simple video conversion using C++ and ffmpeg.

  


  When trying to convert a video using FFmpeg, calling avcodec_open2 fails with the code "-22" which seems to be an "Invalid Argument"-error.

  


  I can't figure out why it fails, nor what the invalid argument is. In the following snippet I create the output codec and pass its context the information from the source (code further down below).

  


  The check for the "outputCodec" works and does not throw an error. As far as I know an "AVDictionary"-argument is optional. So I guess the reason for the error is the context.

  


      const AVCodec* outputCodec = avcodec_find_encoder_by_name(codecName.c_str());

    if (!outputCodec)
    {
        std::cout << "Zielformat-Codec nicht gefunden" << std::endl;
        return -1;
    }

    AVCodecContext* outputCodecContext = avcodec_alloc_context3(outputCodec);
    outputCodecContext->bit_rate = bitRate;
    outputCodecContext->width = inputCodecContext->width;
    outputCodecContext->height = inputCodecContext->height;
    outputCodecContext->pix_fmt = outputCodec->pix_fmts[0];
    outputCodecContext->time_base = inputCodecContext->time_base;

    **int errorCode = avcodec_open2(outputCodecContext, outputCodec, NULL); //THIS RETURNS -22**
    if (errorCode != 0)
    {
        std::cout << "Fehler beim Öffnen des Zielformat-Codecs" << std::endl;
        return -1;
    }


  


  Here is the code for getting the input file & context :

  


      std::string inputFilename = "input_video.mp4";
    std::string outputFilename = "output.avi";
    std::string codecName = "mpeg4";
    int bitRate = 400000;

    AVFormatContext* inputFormatContext = NULL;
    if (avformat_open_input(&inputFormatContext, inputFilename.c_str(), NULL, NULL) != 0)
    {
        std::cout << "Fehler beim Öffnen der Eingabedatei" << std::endl;
        return -1;
    }

    [Do Video Stream Search)

    AVCodecParameters* inputCodecParameters = inputFormatContext->streams[videoStreamIndex]->codecpar;
    const AVCodec* inputCodec = avcodec_find_decoder(inputCodecParameters->codec_id);
    AVCodecContext* inputCodecContext = avcodec_alloc_context3(inputCodec);
    if (avcodec_parameters_to_context(inputCodecContext, inputCodecParameters) != 0)
    {
        std::cout << "Fehler beim Setzen des Eingabecodecs" << std::endl;
        return -1;
    }
    if (avcodec_open2(inputCodecContext, inputCodec, NULL) != 0)
    {
        std::cout << "Fehler beim Öffnen des Eingabecodecs" << std::endl;
        return -1;
    }


  


  The purpose was simply to get started with ffmpeg in an own C++ project.

  


  If it is of any need, I downloaded the ffmpeg libs from here. I used the gpl shared ones. The architecture is win x64. I referenced them through the project properties (additional libraries and so on).

  


  I tried to convert a .mp4 video to an .avi video with an "mpeg4" compression.
I also tried other compressions like "libx264" but none worked.

  


  I searched for the problem on stackoverflow but could not find the exact same problem. While its purpose is different this post is about the same error code when calling avcodec_open2. But its solution is not working for me. I inspected the watch for the "outputContext" while running the code and the codec_id, codec_type and format is set. I use the time_base from the input file. According to my understanding, this should be equal to the source. So I can not find out what I am missing.

  


  Thanks in advance and sorry for my english.

  


  And for completion, here is the whole method :

  


  int TestConvert()
{
    std::string inputFilename = "input_video.mp4";
    std::string outputFilename = "output.avi";
    std::string codecName = "mpeg4";
    int bitRate = 400000;

    AVFormatContext* inputFormatContext = NULL;
    if (avformat_open_input(&inputFormatContext, inputFilename.c_str(), NULL, NULL) != 0)
    {
        std::cout << "Fehler beim Öffnen der Eingabedatei" << std::endl;
        return -1;
    }

    int videoStreamIndex = -1;
    for (unsigned int i = 0; i < inputFormatContext->nb_streams; i++)
    {
        if (inputFormatContext->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
        {
            videoStreamIndex = i;
            break;
        }
    }

    AVCodecParameters* inputCodecParameters = inputFormatContext->streams[videoStreamIndex]->codecpar;
    const AVCodec* inputCodec = avcodec_find_decoder(inputCodecParameters->codec_id);
    AVCodecContext* inputCodecContext = avcodec_alloc_context3(inputCodec);
    if (avcodec_parameters_to_context(inputCodecContext, inputCodecParameters) != 0)
    {
        std::cout << "Fehler beim Setzen des Eingabecodecs" << std::endl;
        return -1;
    }
    if (avcodec_open2(inputCodecContext, inputCodec, NULL) != 0)
    {
        std::cout << "Fehler beim Öffnen des Eingabecodecs" << std::endl;
        return -1;
    }

    const AVCodec* outputCodec = avcodec_find_encoder_by_name(codecName.c_str());

    if (!outputCodec)
    {
        std::cout << "Zielformat-Codec nicht gefunden" << std::endl;
        return -1;
    }

    AVCodecContext* outputCodecContext = avcodec_alloc_context3(outputCodec);
    outputCodecContext->bit_rate = bitRate;
    outputCodecContext->width = inputCodecContext->width;
    outputCodecContext->height = inputCodecContext->height;
    outputCodecContext->pix_fmt = outputCodec->pix_fmts[0];
    outputCodecContext->time_base = inputCodecContext->time_base;

    int errorCode = avcodec_open2(outputCodecContext, outputCodec, NULL);
    if (errorCode != 0)
    {
        std::cout << "Fehler beim Öffnen des Zielformat-Codecs" << std::endl;
        return -1;
    }

    AVFormatContext* outputFormatContext = NULL;
    if (avformat_alloc_output_context2(&outputFormatContext, NULL, NULL, outputFilename.c_str()) != 0)
    {
        std::cout << "Fehler beim Erstellen des Ausgabe-Formats" << std::endl;
        return -1;
    }

    AVStream* outputVideoStream = avformat_new_stream(outputFormatContext, outputCodec);
    if (outputVideoStream == NULL)
    {
        std::cout << "Fehler beim Hinzufügen des Video-Streams zum Ausgabe-Format" << std::endl;
        return -1;
    }
    outputVideoStream->id = outputFormatContext->nb_streams - 1;
    AVCodecParameters* outputCodecParameters = outputVideoStream->codecpar;
    if (avcodec_parameters_from_context(outputCodecParameters, outputCodecContext) != 0)
    {
        std::cout << "Fehler beim Setzen des Ausgabe-Codecs" << std::endl;
        return -1;
    }

    if (!(outputFormatContext->oformat->flags & AVFMT_NOFILE))
    {
        if (avio_open(&outputFormatContext->pb, outputFilename.c_str(), AVIO_FLAG_WRITE) != 0)
        {
            std::cout << "Fehler beim Öffnen der Ausgabedatei" << std::endl;
            return -1;
        }
    }

    if (avformat_write_header(outputFormatContext, NULL) != 0)
    {
        std::cout << "Fehler beim Schreiben des Ausgabe-Formats in die Ausgabedatei" << std::endl;
        return -1;
    }

    AVPacket packet;
    int response;
    AVFrame* frame = av_frame_alloc();
    AVFrame* outputFrame = av_frame_alloc();
    while (av_read_frame(inputFormatContext, &packet) == 0)
    {
        if (packet.stream_index == videoStreamIndex)
        {
            response = avcodec_send_packet(inputCodecContext, &packet);
            while (response >= 0)
            {
                response = avcodec_receive_frame(inputCodecContext, frame);
                if (response == AVERROR(EAGAIN) || response == AVERROR_EOF)
                {
                    break;
                }
                else if (response < 0)
                {
                    std::cout << "Fehler beim Dekodieren des Video-Pakets" << std::endl;
                    return -1;
                }

                struct SwsContext* swsContext = sws_getContext(inputCodecContext->width, inputCodecContext->height, inputCodecContext->pix_fmt, outputCodecContext->width, outputCodecContext->height, outputCodecContext->pix_fmt, SWS_BILINEAR, NULL, NULL, NULL); if (!swsContext)
                {
                    std::cout << "Fehler beim Erstellen des SwsContext" << std::endl;
                    return -1;
                }
                sws_scale(swsContext, frame->data, frame->linesize, 0, inputCodecContext->height, outputFrame->data, outputFrame->linesize);
                sws_freeContext(swsContext);

                outputFrame->pts = frame->pts;
                outputFrame->pkt_dts = frame->pkt_dts;
                //outputFrame->pkt_duration = frame->pkt_duration;
                response = avcodec_send_frame(outputCodecContext, outputFrame);
                while (response >= 0)
                {
                    response = avcodec_receive_packet(outputCodecContext, &packet);
                    if (response == AVERROR(EAGAIN) || response == AVERROR_EOF)
                    {
                        break;
                    }
                    else if (response < 0)
                    {
                        std::cout << "Fehler beim Kodieren des Ausgabe-Frames" << std::endl;
                        return -1;
                    }

                    packet.stream_index = outputVideoStream->id;
                    av_packet_rescale_ts(&packet, outputCodecContext->time_base, outputVideoStream->time_base);
                    if (av_interleaved_write_frame(outputFormatContext, &packet) != 0)
                    {
                        std::cout << "Fehler beim Schreiben des Ausgabe-Pakets" << std::endl;
                        return -1;
                    }
                    av_packet_unref(&packet);
                }
            }
        }
        av_packet_unref(&packet);
    }

    av_write_trailer(outputFormatContext);
    avcodec_free_context(&inputCodecContext);
    avcodec_free_context(&outputCodecContext);
    avformat_close_input(&inputFormatContext);
    avformat_free_context(inputFormatContext);
    avformat_free_context(outputFormatContext);
    av_frame_free(&frame);
    av_frame_free(&outputFrame);

    return 0;

}


  


• 10 Matomo Features You Possibly Didn’t Know About

  28 octobre 2022, par Erin

  Most users know Matomo as the privacy-focussed web analytics tool with data accuracy, superior to Google Analytics. 

  And we’re thrilled to be that — and more ! 

  At Matomo, our underlying product vision is to provide a full stack of accurate, user-friendly and privacy-mindful online marketing tools. 

  Over the years, we’ve expanded beyond baseline website statistics. Matomo Cloud users also get to benefit from additional powerful tools for audience segmentation, conversion optimisation, advanced event tracking and more. 

  Here are the top 10 advanced Matomo features you wish you knew about earlier (but won’t stop using now !). 

  Funnels

  At first glance, most customer journeys look sporadic. But every marketer will tell you that there is a method to almost every users’ madness. Or more precisely — there’s a method you can use to guide users towards conversions. 

  That’s called a customer journey — a schematic set of steps and actions people complete from developing awareness and interest in your solution to consideration and finally conversion.

  On average, 8 touchpoints are required to turn a prospect into a customer. Though the number can be significantly bigger in B2B sales and smaller for B2C Ecommerce websites. 

  With the Funnels feature, you can first map all the on-site touchpoints (desired actions) for different types of customers. Then examine the results you’re getting as prospects move through these checkbox steps.

  Funnel reports provide :

  • High-level metrics such as “Funnel conversion rate”, “Number of funnel conversions”, “Number of funnel entries”. 
  • Drilled-down reports for each funnel and each tracked action within it. This way you can track the success rates of each step and estimate their contribution to the cumulative effect.

  Segmented funnel reports for specific user cohorts (with Matomo Segmentation enabled).

  

  What makes funnels so fun (pun intended) ? The variety of use cases and configurations ! 

  You can build funnels to track conversion rates for :

  • Newsletter subscriptions
  • Job board applications 
  • Checkout or payment 
  • Product landing pages
  • Seasonal promo campaigns

  …. And pretty much any other page where users must complete a meaningful action. So go test this out. 

  Form Analytics

  On-site forms are a 101 tactic for lead generation. For most service businesses, a “contact request” or a “booking inquiry” submission means a new lead in your pipeline. 

  That said : the average on-site form conversion rates across industries stand at below 50% : 

  • Property – 37% 
  • Telecoms – 40%
  • Software — 46.83%

  That’s not bad, but it could be better. If only you could figure out why people abandon your forms….

  Oh wait, Matomo Form Analytics can supply you with answers. Form Analytics provide real-time information on key form metrics — total views, starter rate, submitter rate, conversions and more.

  

  Separately the average form hesitation time is also provided (in other words, the time a user contemplates if filling in a form is worth the effort). Plus, Matomo also tracks the time spent on form submission.

  You can review : 

  • Top drop-off fields – to understand where you are losing prospects. These fields should either be removed or simplified (e.g., with a dropdown menu) to increase conversions.
  • Most corrected-field – this will provide a clear indication of where your prospects are struggling with a form. Providing help text can simplify the process and increase conversions. 
  • Unesserary fields – with this metric, you’ll know which optional fields your leads aren’t interested in filling in and can remove them to help drive conversions. 

  With Form Analytics, you’ll be able to boost conversions and create a better on-site experience with accurate user data. 

  A/B testing

  Marketing is both an art and a science. A/B testing (or split testing) helps you statistically verify which creative ideas perform better. 

  A good conversion rate optimisation (CRO) practice is to test different elements and to do so often to find your top contenders.

  What can you split test ? Loads of things :

  • Page slogans and call-to-actions 
  • Button or submission form placements
  • Different landing page designs and layouts
  • Seasonal promo offers and banners
  • Pricing information 
  • Customer testimonial placements 

  More times than not, those small changes in page design or copy can lead to a double-digit lift in conversion rates. Accounting software Sage saw a 30% traffic boost after changing the homepage layout, copy and CTAs based on split test data. Depositphotos, in turn, got a 9.32% increase in account registration rate (CR) after testing a timed pop-up registration form. 

  The wrinkle ? A/B testing software isn’t exactly affordable, with tools averaging $119 – $1,995 per month. Plus, you then have to integrate a third-party tool with your website analytics for proper attribution — and this can get messy.

  Matomo saves you the hassle in both cases. An A/B testing tool is part of your Cloud subscription and plays nicely with other features — goal tracking, heatmaps, historic visitor profiles and more. 

  You can run split tests with Matomo on your websites or mobile apps — and find out if version A, B, C or D is the top performer. 

  

  Advertising Conversion Exports

  A well-executed search marketing or banner remarketing campaign can drive heaps of traffic to your website. But the big question is : How much of it will convert ?

  The AdTech industry has a major problem with proper attribution and, because of it, with ad fraud. 

  Globally, digital ad fraud will cost advertisers a hefty $8 billion by the end of 2022. That’s when another $74 million in ad budgets get wasted per quarter. 

  The reasons for ad budget waste may vary, but they often have a common denominator : lack of reliable conversion tracking data.

  Matomo helps you get a better sense of how you spend your cents with Advertising Conversion Reports. Unlike other MarTech analytics tools, you don’t need to embed any third-party advertising network trackers into your website or compromise user privacy.

  Instead, you can easily export accurate conversion data from Matomo (either manually via a CSV file or automated with an HTTPS link) into your Google Ads, Microsoft Advertising or Yandex Ads for cross-validation. This way you can get an objective view of the performance of different campaigns and optimise your budget allocations accordingly. 

  Find out more about tracking ad campaigns with Matomo.

  Matomo Tag Manager

  The marketing technology landscape is close to crossing 10,000 different solutions. Cross-platform advertising trackers and all sorts of customer data management tools comprise the bulk of that growing stack. 

  Remember : Each new tool embed adds extra “weight” to your web page. More tracking scripts equal slower page loading speed — and more frustration for your users. Likewise, extra embeds often means dialling up the developer (which takes time). Or tinkering with the site code yourself (which can result in errors and still raise the need to call a developer). 

  With Tag Manager, you can easily generate tags for :

  • Custom analytics reports 
  • Newsletter signups
  • Affiliates 
  • Form submission tracking 
  • Exit popups and surveys
  • Ads and more

  With Matomo Tag Manager, you can monitor, update or delete everything from one convenient interface. Finally, you can programme custom triggers — conditions when the tag gets activated — and specify data points (variables) it should collect. The latter is a great choice for staying privacy-focused and excluding any sensitive user information from processing. 

  With our tag management system (TMS), no rogue tags will mess up your analytics or conversion tracking. 

  Session recordings

  User experience (UX) plays a pivotal role in your conversion rates. 

  A five-year McKinsey study of 300 publicly listed companies found that companies with strong design practices have 32 percentage points higher revenue growth than their peers. 

  But what makes up a great website design and browsing experience ? Veteran UX designers name seven qualities :

  

  To figure out if your website meets all these criteria, you can use Session Recording — a tool for recording how users interact with your website. 

  By observing clicks, mouse moves, scrolls and form interactions you can determine problematic website design areas such as poor header navigation, subpar button placements or “boring” blocks of text. 

  Such observational studies are a huge part of the UX research process because they provide unbiased data on interaction. Or as Nielsen Norman Group puts it :

  “The way to get user data boils down to the basic rules of usability :

  • Watch what people actually do.
  • Do not believe what people say they do.
  • Definitely don’t believe what people predict they may do in the future.” 

  Most user behaviour analytics tools sell such functionality for a fee. With Matomo Cloud, this feature is included in your subscription. 

  Heatmaps

  While Session Replays provide qualitative insights, Heatmaps supply you with first-hand qualitative insights. Instead of individual user browsing sessions, you get consolidated data on where they click and how they scroll through your website. 

  

  Heatmaps are another favourite among UX designers and their CRO peers because you can :

  • Validate earlier design decisions around information architecture, page layout, button placements and so on. 
  • Develop new design hypotheses based on stats and then translate them into website design improvements. 
  • Identify distractive no-click elements that confuse users and remove them to improve conversions. 
  • Locate problematic user interface (UI) areas on specific devices or operating systems and improve them for a seamless experience.

  To get even more granular results, you can apply up to 100 Matomo segments to drill down on specific user groups, geographies or devices. 

  This way you can make data-based decisions for A/B testing, updating or redesigning your website pages. 

  Custom Alerts

  When it comes to your website, you don’t want to miss anything big — be it your biggest sales day or a sudden nosedive in traffic. 

  That’s when Custom Alerts come in handy. 

  

  With a few clicks, you can set up email or text-based alerts about important website metrics. Once you hit that metric, Matomo will send a ping. 

  You can also set different types of Custom Alerts for your teams. For example, your website administrator can get alerted about critical technical performance issues such as a sudden spike in traffic. It can indicate a DDoS attack (in the worst case) — and timely resolution is crucial here. Or suggest that your website is going viral and you might need to provision extra computing resources to ensure optimal site performance.

  Your sales team, in turn, can get alerted about new form submissions, so that they can quickly move on to lead scoring and subsequent follow-ups. 

  Use cases are plentiful with this feature. 

  Custom Dashboards and Reports

  Did you know you can get a personalised view of the main Matomo dashboards ? 

  By design, we made different website stats available as separate widgets. Hence, you can cherry-pick which stats get a prominent spot. Moreover, you can create and embed custom widgets into your Matomo dashboard to display third-party insights (e.g., POS data).

  Set up custom dashboard views for different teams, business stakeholders or clients to keep them in the loop on relevant website metrics. 

  

  Custom Reports feature, in turn, lets you slice and dice your traffic analytics the way you please. You can combine up to three different data dimensions per report and then add any number of supported metrics to get a personalised analytics report.

  For example, to zoom in on your website performance in a specific target market you can apply “location” (e.g., Germany) and “action type” (e.g., app downloads) dimensions and then get segmented data on metrics such as total visits, conversion rates, revenue and more. 

  Get to know even more ways to customise Matomo deployment.

  Roll Up Report

  Need to get aggregated traffic analytics from multiple web properties, but not ready to pay $150K per year for Google Analytics 360 for that ?

  We’ve got you with Roll-Up Reporting. You can get a 360-degree view into important KPIs like global revenue, conversion rates or form performance across multiple websites, online stores, mobile apps and even Intranet properties.

  

  Setting up this feature takes minutes, but saves you hours on manually exporting and cross-mapping data from different web analytics tools. 

  Channel all those saved hours into more productive things like increasing your conversion rates or boosting user engagement. 

  Avoid Marketing Tool Sprawl with Matomo 

  With Matomo as your website analytics and conversion optimisation app, you don’t need to switch between different systems, interfaces or have multiple tracking codes embedded on your site.

  And you don’t need to cultivate a disparate (and expensive !) MarTech tool stack — and then figure out if each of your tools is compliant with global privacy laws.

  All the tools you need are conveniently housed under one roof. 

  Want to learn more about Matomo features ? Check out product training videos next ! 

  21 day free trial. No credit card required.

  Your email address
  

  Your website address
  

  Your Analytics subdomain will be .matomo.cloud (change)

  Choose your analytics subdomain
  

  .matomo.cloud

  I hereby accept the terms and conditions and the data processing agreement (DPA).

  
  Your information will be used to create an account on our cloud service. For more information please consult our privacy policy.
  
  

  » Start improving your websites now

  

• Android, AOSP tree, external project (ffmpeg) is built for AMD64

  22 avril 2015, par kagali-san

  I’m trying to build ffmpeg4android on current AOSP tree (from /external), which is lunch-configured to aosp_arm-eng and set to PLATFORM_VERSION=4.2.

  Resulting files are generated for AMD64 (host native) architecture, even though the major rest of tree is built (as expected) ARM :

  readelf -a android/aosp_arm-eng/ffplay|egrep "Class :|Machine :"
  Class : ELF64 Machine :
  Advanced Micro Devices X86-64

  versus

  readelf -a
  aosp/out/target/product/generic/symbols/system/lib/libril.so | egrep "Class :|Machine :"

  Class : ELF32
  Machine : ARM

  ---

  I will probably switch to other ways of getting ffmpeg-arm (presumably the one described here) ; the reason of asking this question is to understand, at which build stage does cross-compilation environment breaks.